ReportNo. 15158-YEM Republic of Assessmzent Public Disclosure Authorized

June26, 1996

Middl(e I st I IlImmi RemMIrcesD)ivi,(fln ( o LIltrV DX I IeeItII Middle East mindNorthi Africa Retgion Public Disclosure Authorized Public Disclosure Authorized

Public Disclosure Authorized Document of the World Bank ClURRENCY EQuIVALENTS

Currency Unit = Yemeni Rial (YR) YR 1.0 = 100 Fils

Exchange Rate as of.June 1, 1996

Official exchange rate: YR II 5/US$ Parallel exchanigerate: YR I1 5/1US$

Yemen Fiscal Year

January I - December 3 1

LIST OF ACRONYMS

ARI Acute Respiratory Infectionis DHS Demographilc and Maternal and Child Health Survey EPI Expanded Program of Iin Inllnizationis FAO Food and Agriculture Organization GDP Gross Domestic Product GNP Gross National Product HBS-92 Household Budget Survey, 1992 LCCD Local Counicils for Cooperative Development LDA Local Developmenlt Association MOSL Ministry of Social Affairs, Social Insiuranice,and Labor NGO NongoverinmenltalOrgan ization OECD Organ isation for Economic Co-operation aiid Development PDRY Peoples' Democratic Republic of Yemen TFR Total Fertility Rate UNDP lUnited Nations Development Programine UNHCR ULited Nations High Commission for Reftugees WDR World Development Report YAR Yemen Arab Republic YR Yeienl Rial REPUBLIC OF YEMEN POVERTY ASSESSMENT

TABLEOF CONTENTS

i EXECUTIVESUMMARY ...... I THE POVERTY PROFILE ...... ,. 2 THE EXTENTAND DEPTHOF POVERTY...... 3 A DESCRIPTIONOF THE POOR...... 5 Sectoral and Regional Comparison of Poverty ...... 6 Education ...... 8 Labor Force Activities ...... 9 Summary Household Charactei istics ...... 10 Groups of Poor ...... 10 AN OVERVIEW: REGRESSIONANALYSIS ...... POOR ...... 12 SHORT-TERM IMPLICATIONS OF ECONOMIC POLICIES ON THE ...... 12 THE GOVERNMENT'SSOCIAL SAFETY NET: UNIVERSALSUBSIDIES 14 WHEATSUBSIDY ...... 14 Major Issues ...... 16 Options ...... 17 SOCIAL SAFETYNETS: PUBLICOR PRIVATEINITIATIVE? ...... 17 Major Issues ...... 20 Options ...... POOR...... 22 MEDIUM-TERM IMPLICATIONS OF ECONOMIC POLICIES ON THE 22 THE FOUNDATIONSFOR ECONOMICDEVELOPMENT ...... 23 Major Issues ...... 26 Options ...... ON THE POOR ...... 29 LONG-TERM IMPLICATIONS OF SOCIAL SECTOR POLICIES 30 T HEALHE THSECTOR ...... 32 Major Issues ...... 34 Options ...... 36 THE EDUCATIONSECTOR ...... 36 Major Issues...... 41 Options ...... 43 CONCLUSION ...... 45 ANNEXES ...... 59 REFERENCES ......

Peter Heywood, Arun Joshi, Donald This report is the work of Sarosh Sattar (Task Manager), Nabil Alnawwab (ESCWA), by Bahjat Achikbache, Jacqueline Baptist, Mitchell, Hussein Shakhatreh, and Kamar Yousuf. Additional support was provided visited Sana'a during September 1995 and were Sharon Beatty, Tahir Qassim, and Mona Said. Most members of the team was assisted by the generous and industrious received under the auspices of the Ministry of Planning. In Sana'a the mission Office, Ministry of Finance, Ministry of counterparts throughout the Yemeni govemment, in particular the Central Statistical Chu (IMF) and Martin Ravallion were peer Education, Ministry of Health, and Ministry of Labor and Social Affairs. Ke-young reviewers.

EXECUTIVESUMMARY

1. Poverty is widespread in Yemen, affecting one out of every five persons. By definition, the poor's expenditures are below socially acceptable norms, but in Yemen the poor also suffer more than the nonpoor from ill health and lower educational attainment. This raises the concern of whether public policies and expenditures are adequate for promoting economic and social development since such a large share of the population is excluded from participating in the benefits.

2. The options available to the Government for assisting the poor are limited by two characteristics of Yemen. First, the poor live mostly in rural areas (81 percent of all poor) which makes it difficult to reach them through public assistance programs. Second, public institutions are weak and the Gevernment must choose options which are easily implementable and involve partners from the private and nonprofit sectors in order to alleviate poverty.

3. The strategy for Yemen has three components: (i) policies which encourage broad-hased economic growth, (ii) increased provision of social services, and (iii) development of the iiongoverninental sector. The main steps the Government must take are to (a) phase out the universal subsidy program, (b) improve the effectiveness of public expenditures, and (c) revisit the strategies in the health and education sectors. Macroeconomic stability is a prerequisite for the success of these actions.

OVERVIEW OF THE POVERTY PROFILE

4. Yemen's growth rate of 3.3 percent per annum is one of the highest in the world. Among the countries in the Middle East and North Africa, Yemen has the lowest life expectancy (51 years), the lowest adult literacy rate (38 percent), the highest fertility rate (TFR 7.5), and the highest infant mortality rate (11.7 percent of live births).

5. Poverty in Yemen is measured by establishing minimum food and nonfood expenditure requirements for each governorate and rural and urban areas. Analysis of a recent household survey indicates that over half of Yemen's poor live in four governorates (Sana'a, Taiz, Ibb, and Dhamar) and an additional 21 percent live in Hodeidah, Hadramout, and Al-Beida. The greatest concentration of poverty is in Dhamar, Al-Beida, and Al-Mahwit where one-third of the governorates' population live below the poverty line.

6. Though poverty is widespread in Yemen, it is primarily a rural phenomenon. Most of the population lives in the rural areas. In 1995 .he population was 15.3 million of which about 0.6 million were urban poor (4 percent) and 2.4 rmillion(15 percent) rural poor.1

7. The labor force participation rates of the poor and nonpoor are similar at about 53 percent though poor men join the labor force at a younger age and exit at an older age than nonpoor men. In 1992 accoiding to the Household Budget Survey an estimated 9 percent of the poor were unemployed compared to 7 percent of the nonpoor. The (employed) poor work mostly (82 percent) in the private sector either as self-employed or employees compared to 76

The total populationfor 1995 is a World Bank estimate. The proportion of the rural population is estimated at 80 percent of total population using census data from 1986 and 1988. Updated informationfrom the 1994 censusis not yet available. - 11 - percent of the nonpoor. The public sector employs a significant share of the labor force (25 percent). In 1992, 13 percent of all public sector employees fell below the poverty line compared to 18 percent of private sector workers.

8. Poor persons are less likely to be literate than the nonpoor in Yemen: 66 percent of the poor are illiterate compared to 59 percent of the nonpoor. Analysis of the Household Budget Survey (1992) shows that (i) adult men are more likely to be literate - poor or nonpoor - than women by a factor of four times on average; (ii) literacy among younger men - whether poor or nonpoor - is greater than among older men which implies that the Government has been successful in delivering education services to the male population; and (iii) literacy and school enrollments vary greatly among governorates with Aden having the largest proportion of literate adults (72 percent) and Hajjah the lowest (17 percent).

GOVERNMENTACTIONS TO REDUCEPOVERTY

9. Since the Government has initiated a stabilization and structural adjustment program, the economic policy foundation presents an opportunity for improving public expenditures and sector strategies. Over the next several years, Yemen can reduce poverty if it improves the environment for private investment, foilows sensible fiscal and monetary policies, and wisely uses its oil revenues (estimated at about 22 percent of GDP in 1995). However, if the Government continues to use this income to expand public hiring or transfers to the population, then short-term consumption will off-set the possibility for long-term development.

SHORT-TERMIMPLICATIONS OF ECONOMIC POLICIES

10. The first step for the Government is to redirect its expenditures away from its expensive and ineffective safety net program of universal subsidies which costs an estimated 16 percent of GDP in 1994. The wheat and flour subsidy is one of the major components of this program and costs about 5 percent of GDP or 25 percent of the Government's oil revenue. In 1995 only a maximum of $65 million of the $324 million subsidy is estimated to have reached the poorest 20 percent of the population.

II. In order to eliminate poverty the Government would need to transfer about 5 percent of GDP directly to the poor. Though the amount to be transferred is affordable (since it is equal to the wheat subsidy), such a program could not be implemented in Yemen without large leakages to the nonpoor. Since the Government does not have the capacity to target the poor, the current universal subsidy program should not be replaced by a publicly administered or financed transfer program.

12. The poor can best be helped by encouraging the development of the nongovernmental sector which would include both nonprofit organizations and self-help community organizations. NGOs can provide alternative and cost-effective methods of reaching the poor. Though there are only about fifteen organizations which are actively involved in delivering reasonable quality social services to the population, the nongovernmental sector is growing. - iii -

13. The development of the sector can be encouraged in two ways. First, the Government can strengthen the sector by (i) removing obstacles to the expansion of private activity, especially regulations which stifle initiatives; (ii) setting transparent criteria in the Ministry of Social Affairs, Social Insurauce, and Labor (MOSL) for allocating funds to NGOs; (iii) providing incentives to welfare organizations to help the poor; (iv) facilitating increased donor funding of NGOs, (v) facilitating the provision of technical and management training for NGO staff, and (vi) disengaging itself from involvement in Local Councils of Community Development

14. Second, the proposed Social Fund project (supported by the World Bank) would increase the financial resources allocated to NGOs or local communitv groups which assist the poor. The MOSL budget for cooperative societies (including NGOs) was YR 27 million in 1994 (US$ 0.4 million) which was divided among 550 organizations.

MEDIUM-TERMIMPLICATIONS OF ECONOMIC POLICI ES

15. Second, the Government should focus more heavily on promoting those policies which increase economic growth. For example, if the economy grows by 6 percent rate per annum, poverty will be cut in half within the next 10 years (assuming that income distribution does not change). Two areas that the Government needs to focus on are improving the distribution of public expenditures and improving the profitability of the agriculture sector in the medium-term.

16. The allocation of resources between recurrent and capital expenditures and among different recurrent categories may inhibit the effectiveness of public intervention in the economy. The Government may want to reconsider (i) the recent cuts in material expenditures (Chapter 2 of the budget) which have fallen to 8 percent of public expenditures; (ii) the high pace of public sector hiring estimated at 15 percent per anium during 1990 - 1994; and (iii) the low level of public domestic investmenit in "development" activities (about 7 percent in 1995).

17. The agriculture sector is a major employer of labor (58 percent of the labor force), especially low income groups. Lessons from other countries have shown that can be significantly reduced through higher agricultural yields and development expenditures in the rural sector. Specific actions the Government can undertake whicih will improve the profitability of the agriculture sector and thus rural incomes are: (i) elimination of the wheat and flour subsidy which will remove the bias against farmiers and the domestic production of grains; (ii) improvement of investments in physical and social infrastructure in the rural sector; (iii) instituting policies and incentives which will reverse rural practices causing the depletion of water resources and deforestation such as the monopolization in rural areas of water resources by better off farmers, and (iv) improvement of research and extension services.

LONG-TERM IMPLICATIONS OF SOCIAL SECTOR POLICIES

1 8. Finally, an important element for a successful poverty alleviation strategy in Yemen is human resources development. Factors which have had an inhibiting effect on the expansion of social services include the low urbanization rate and the Government's role as the major provider of social services.

19. Though there are many obstacles to wider coverage of the population, provision of services could be expanded under the proper incentive regime. To improve living conditions of the poor and nonpoor, the Government should (i) ensure the provision of a basic health package - Iv - to the population and (ii) increase basic school enrollment. Though these goals appear straightforward, they imply changes in the Government's global strategies in health and education and new financing and delivery mechanisms.

20. Public health expenditures in Yemen were 1.7 percent of GDP during 1990 - 1994 compared to 2.5 percent for the Middle East and sub-Saharan Africa regions. Yemen's infant mortality rate (a good indicator of overall health conditions in a country) is high at 117 even when compared to the average for low income countries (64). To improve health conditions, the Government needs to ensure the provision of essential health services to the population and improve female literacy, availability of safe water, and sanitation.

21. Since health conditions are generally low for both the poor and nonpoor, the Government will need to improve health services for the whole population. First, it shiould formulate a clear health policy which specifically addresses how basic public health and clinical services of acceptable quality will be delivered to the rural sector and to low income groups. Second, access to basic health services (including family planning and maternal child health services) needs to be improved using new methods of service delivery including autonomy of public sector providers, the private sector, and nongovernmental organizations. Third, the Government should formulate a new pharmaceuticals policy which removes the state monopoly on procurement.

22. Public expenditures on education are 15 percent of total expenditures and 5 percent of GDP. Yemen's literacy rate (38 percent) is significantly below the average of low income countries (69 percent). Primary and secondary school enrollment rates of 76 percent and 31 percent respectivt'ly are well below those of developing countries. Differences in enrollment rates of the poor and nonpoor are not large. rFl major reaso-Asfor Yemen's low enrollment rates are that relati,ely few girls and rural residents enroll in schools. For example, for every 100 boys enrolled in schcol, tht;re are only 39 girls.

23. The Government's priorities in the education sector should be to increase enrollment rates of girls and rural residents. Possible interventions to increase female enrollments are (i) construct culturally appropriate facilities with toilets for girls and possibly with walls around girls' classrooms; (ii) increase the number of female schools or have an additional shift only for girls; (iii) launch information campaigns that engage community, religious, and civic leaders; (iv) provide (financial) incentives for female teachers; (v) adjust the school calendar and schedule to accommodate household child labor requirements; and (vi) provide families with incentives for sending their girls to school (such as paying for female student's school materials and uniforms).

24. The Government will need to do more analysis of why rural enrollment rates are so low. If it is due to the shortage of schools then the Government must find ways of increasing the delivery of services such as importing teachers, providing teachers incentives to work in rural areas, or financing the private sector to deliver services in the rural areas.

25. Additional actions the Government should take are to (i) establish a strategy and school enrollment targets for the next five years; (ii) increase community involvement in school construction, financing of teachers, and quality control of teachers; (iii) find ways of decreasing the private cost of education for poo. children such as suspending the school uniform requirement; and (iv) establishing a data and information system. 1. THE POVERTY PROFILE

COUNTRY BACKGROUND

1.1 The Republic of Yemen is a young nation-state formed in 1990 through the unification of the Yemen Arab Republic (YAR) and the People's Democratic Republic of Yemen (PDRY). The YAR was established in 1963 at the end of the Imamate which followed the Ottoman occupation (1849 - 1919). The PDRY achieved independence from colonial rule in 1970. The British occupation of southern Yemen began with the annexation of Perim, an island in the Arabian Sea, followed by the conquest of Aden in 1839.

1.2 Yemen's population was 15.3 million in 1995.1 The growth rate of 3.3 percent per annum is one of the highest in the world. Among the countries in the Middle East and North Africa, Yemen has the lowest life expectancy (51 years), the lowest adult literacy rate (38 percent), the highest fertility rate (TFR 7.5), and the highest infant mortality rate (11.7 percent of live births).

1.3 Yemen is a low income country with an estimated GDP of US $314 per capita in 1995. The three characteristics which set Yemen apart from other lower income countries are (i) modest oil reserves, (ii) production and consumption of qat , and (iii) an unraveling centralized economy in the former governorates of the PDRY. Oil income accruing to the Government is estimated at 23 percent of GDP in 1994-1995. Qat plays an important role in Yemen's economy significantly affecting the agriculture sector, labor productivity, and allocation of household resources. National unification led to the dismantling of the PDRY's centralized and socialist economic structure without providing an environment conducive to the private sector growth.

1.4 Two major shocks occurred in 1991 whose impact continues to affect Yemen's economy. First, the Gulf crisis led to the repatriation of about one million Yemeni's (8 percent of the population) and the loss of substantial worker remittances. The country not only lost an important source of income, but also the size of the labor force grew without a corresponding increase in employment opportunities. Second, the unification of YAR and PDRY has led to a large civil service, an unstable political situation which resulted in a civil war in 1994, and a difficult law and order situation.

1.5 As a consequence of these shocks, the fiscal deficit increased to 14 percent of GDP in 1994 and inflation averaged 58 percent per annum during 1990 - 1994. The macroeconomic situation deteriorated sharply particularly affecting the urban areas and formal sector investment. The Government has recently taken bold steps to decrease its overspending which has led to a stabilization in prices benefiting all, but especially the poor. The deficit has fallen to 8 percent of GDP and inflation to 48 percent.

Source: WorldBank. Accordingto preliminaryresults of the December1994 census,Yemen's total residentpopulation was 14.6million.

2 Qat is an agricultural crop which is cultivated in areas with high rainfall. Its leaves contain an amphetaminewhich is used mostly by adult males to achieve an altered state of mind. Though it decreasesappetite, it is not physicallyaddictive and does not have any known significanthealth side effects. It is an important part of Yemeni social life although it is seldom consumed outside of Yemen. -2 -

PopulationDistribution

C1 5 - 10 //\

THE EXTENT ANDDEPTHOFYAL MAHARA z > / t ~~~~~~~~HADRAMOUT

AL MA HBW

HODYIDA D A AV

Source:Central Statistical Organization Government of Yemen.

Thepopulation distribution map showsdistribution of Yemen's population by governorate.The denominator is total populationof Yemen.

THE EXTENTAND DEPTH OF POVERTY

1.6 In 1992 an estimated 19 percent of the population lived in poverty which means that 2.6 million persons had insufficient resources to meet their basic food and nonfood requirements. The share of the population in absolute poverty -with total expenditures insufficient to meet even their food needs - was 9 percent. In 1995 the share of the population living in poverty remained constant at 19 percent due to stagnant economic growth.3 The average upper poverty line is YR 9,134 per year per capita or $203. The lower poverty line is YR 6,371 or $142 per capita. (See Annex I for an explanation of poverty line calculations.)

1.7 Though it is difficult to make international comparisons, poverty in Yemen appears to be much higher than in other Middle East countries. For example, the share of the population living in poverty in Egypt and Iran is 6 percent and 9 percent respectively for 1990. Algeria, Tunisia, and Morocco each had less than 3 percent of their population living in poverty.4

This assumesthat the distributionof expendituresduring 1992 - 1995 remained constant. No new informationon expenditureshas been collectedafter 1992. There are two internationalmeasurements of poverty. A person is defined as "poor" if he/she has expenditureswhich are either less than (i) $1 per capita per day in 1985 purchasingpower parity or (ii) expendituresrequired for the consumptionof 2200 caloriesper day per capita. The lower poverty line for Yemen uses the latter measure. Informationprovided above for other Middle East countries employsthe $1 per day measure. - 3 -

Table 1.1: Average Expenditures of Ihe Poor in 1992

Shareof Population AverageExpenditures AverageExpenditures (in percent) (YRlyear/capita) ($/year/capita) AbsolutePoor 9 4,569 119 Poor 19 6,435 162 Nonpoor 81 24,228 540

All Yemen 100 20,885 464

Source: World Bank stafl calculationsbased on HBS-92. Notc: Averageexpenditlres in l!S$ are calculaledat the parallelmarket exchangerate of YR 45/US$.

1.8 Though the poor compriseabout one-fifth of the population,they consumeonly 6 percent of the total private expendituresin Yemen. Figure 1.1 showsthe distribution of expendituresby quintile. The richest 20 percent of the population consumesalmost half of all expenditures. Inequality in Yemen is average for a low income country and for the Middle East region as measuredby the of 0.395.

Figure 1.1: Distribution of Private Expenditures in Yemen

6% 110%

46% . \Poorest 20% 15% .2nd quintile _3rd quintile

0 4th quintile Richest20%

Source: World Bank staff calculationsbased on HBS-92.

A DESCRIPTION OF THE POOR

1.9 This section attempts to answer the question of who are the poor by providing an overview of where they live and what are their characteristics,especially those related to educationand employment. The HouseholdBudget Surveyundertaken in 1992 is the basis for this analysis. It is the only survey which has collected data on householdexpenditures since unification. Expendituresper capita ratherthan incomeare usedto distinguishthe poor from the

A Gini coefficient equal to one indicates perfect inequality and zero implies perfect equality of income or expenditures. Gini coefficients for Algeria, Jordan, Morocco, and Tunisia are 0.39, 0.43, 0.40. and 0.40 respectively. -4 -

BOX 1.]

THEPOVERTYLJNE

The World Bank prefers to use a poverty line provided by the Government since "poverty" is a relative term and a country's citizens can best determine what constitutes an acceptable living standard. In the case of Yemen, the Government did not have a poverty line and requested that the Bank assist in determining an acceptable minimum level of expenditures.

A poverty line was calculated for each of the govemorates and urban and rural locations. The average (weighted) poverty line in 1992 is estimatedat YR 9,134 per year per capita, is the minimum required expenditures to live in Yemen. It includes both a food component and a nonfood component which are estimated to cost YR 6,371 and YR 2,763 respectively.

The food consumption bundle in Yemen is determined using the most recent Food and Agriculture Organization's Food Balance Sheets for the Yemen Arab Republic in 1984 - 1986. The food balance sheets present a comprehensive picture of the pattem of a country's[ food supply during a specified reference period. It shows per caput supply of 50 food products in kilograms per year. This information was combined with prices for 32 food items from 12 urban centers (each from a different governorate) to calculate a food consumption bundle. The Govemment of Yemen does not publish information on rural retail prices of consumer goods, thus to obtain an estimate of the rural poverty line for each govemorate, rural prices are assumed to be 15 percent below urban prices. This assumption is based on observed price differences between rural and urban areas in other countries which range froin 15 to 30 percent. Certain modifications, which are presented in Annex 1, were made due to the limitation of price information.

Nonfood expenditures are difficult to quantify individually because of the lack of adequate consumption and price data. As a result, the average proportion of nonfood to total expenditures around the lower (food) poverty line was used. (See Annex 1.) The nonfood component is estimated at 29 percent of the minimum total expenditure requirements.

The poverty line assumes a consumption of 2200 calories per day. It does not differentiate by age, gender, or activity level. Thus, irrespective of how old an individual is or how great his/her energy expenditure is, a minimum requirementof 2200 calories is assumed.

The high level of inflation and changes in consumer subsidies in Yemen require reevaluating the poverty line every year. For 1994 the poverty line was recalculated using the official consumer price index as well as the individual prices from Statistical Yearbook 1995. Since there was a significant difference, a poverty line based on updated prices was used because it includes more disaggregated data. The 1994 poverty line is estimated at YR 16,488 per capita per yeat.

nonpoor.6 This is a common approach taken in poverty analysis for two reasons. First, reported income is unreliable due to the tendency of people to understate their earnings. Second, consumption is a better indicator of welfare because households can protect their living standards from income fluctuations to some extent.

6 See AnnexI for a discussionon the statisticalbasis for the povertyanalysis. - 5 -

SECTORAL AND REGIONAL COMPARISONOF POVERTY

1.10 Poverty is primarily a rural phenomenon - 81 percent of all poor and 83 percent of the absolute poor live in rural locations. The rural poor's average expenditure per capita is 31 percent below the rural poverty line compared to the urban poor's average expenditure which is 27 percent below the urban poverty line. About 43 percent of all poor live in three governorates (Sana'a,7 Taiz, and lbb) and an additional 28 percent live in Dhamar, Hodeidah,and Hadramout. There are eleven governorates that each have less than 5 percent of the poor. This is due partly to the small size of the governorates' population as seen in Annex II.A.

1.11 The three measurements of the incidence of poverty used are the headcount index, the , and the squared poverty gap index. These measures provide information on the concentration of poverty within locales (or governorates) rather than the distribution of poverty across locales (or governorates).

Table 1.2: Summary of Po;,'rtJ, Indices by Sector

Population Distribution Average Headcount PovertyGap Squared Sector Distribution of Poor exp/capita Index Index PovertyGap of Poor Index l______(in percent) (in percent) (YR/month) (in percent) (in percent) (in percent)

Urban 19.9 19.4 7.465 18.6 5.1 2.2 Rural 80.1 80.6 6,187 19.2 5.9 2.7

All Yemen 100.0 100.0 6,435 19.1 5.7 2.6

Source: World Bankstaff calculationsbased on I IBS-92.

.12 The headcount index measures how many persons in a particular area have expenditures which are equal to or below the poverty line. In Yemen the highest concentration of poverty as measured by the headcount index is in the rural area and in the governorates of Dhamar, Al- Beida, and Mahwit where almost one-third of the population is poor. The govemorates which border Saudi Arabia such as Al-Jawf, Saadah, and Shabwa have the lowest concentration of poverty with less than 7 percent of the population living in poverty. 1.13 The headcount index does not measure the depth of poverty. A person with expenditures slightly below the poverty line is weighted equally with a person with much lower expenditures. This difficulty is overcome by the poverty gap index. It is the additional money the average poor person would have to spend to reach the poverty line, expressed as a percentage of the poverty line. For example, a poverty gap of 10 percent indicates that the average poor person's expenditures are 90 percent of the poverty line. This poverty indicator shows that poverty is more severe in the rural areas and in the governorates of Dhamar, Al-Mahwit, and Sana'a.

7 Sana'a refersto Sana'a governorateand excludesthe capital,Sana'a City 8 For example, though 32 percent of Al-Mahwit's population lives in poverty, only 4 percent of Yemen's poor live in this governorate. This differencebetween distribution and headcountindex is due to differencein governorates'population sizes. Less than 3 percentof Yemen's populationlives in Al-Mahwit. - 6 -

1.14 The squared poverty gap index measures the severity of poverty as well. Though its meaning is less intuitive, it possesses better analytical properties than that of the poverty gap. It is the mean of the squared proportionate poverty gaps. This index shows that the depth of poverty is most severe in the rural sector and the governorates of Dhamar, Al-Mahwit, and Hadramout.

EDUCATION

1.15 Yemen's government only began to devote significant resources towards education a little over twenty years ago. As a result, the level of adult illiteracy of 59 percent is very high by international standards. For example, the average illiteracy rate in other low income countries is 41 percent which is almost one-third less than in Yemen.

1.16 Poor persons are less likely to be literate than the nonpoor in Yemen. Table 1.3 shows that a larger share of the poor have no education than the nonpoor: 66 percent of the poor are illiterate compared to 59 percent of the nonpoor.

1.17 Analysis of the 1lousehold Budget Survey yields three important points about the relationship between poverty and education levels. The first finding is that the correlation between education and poverty is much weaker than between education level and sex. For example, illiteracy among poor and nonpoor adult women is in excess of 80 percent whereas among poor and nonpoor males, the shares are 44 percent and 34 percent respectively. (See Annex II.B).

1.18 The second point is that there is a strong correlation between age and level of education for both the poor and nonpoor, though it is stronger for men than for women. As seen in Figure 1.2, literacy rises with each successive age group indicating that the Government has taken major steps forward in the delivery of education services to the population, especially for boys. It is also apparent though that the inequality of educational attainment is growing between the most privileged group - nonpoor men -and the least privileged - poor women.

Table 1.3: Level of Education by Poor and Nonpoor (18 years and above)

Distribution Incidence EducationLevel All Yemen Poor Nonpoor Poor Nonpoor

Illiterate 60 66 59 19 81 Literate 20 18 20 16 84 Primary 5 5 5 15 85 Intermediate 5 5 5 17 83 Secondary I I 1 15 85 Communitycollege 6 4 6 11 89 Post-secondaryinstitute I I 1 21 80 University 2 <1 2 4 96 Post-graduatedegree <1 <1 <1 24 76

All Yemen 100 100 100

Source: World Bank staff calculationsbased on HBS-92. - 7 -

Figure 1.2: Literacy by Age. Sex, and Poverty

80 57%ofpoor31-35year | 70 old malesare literate -60___ ~50t03uSO-S_ ' 40co _ i 1j | L . . Male nonpoorX 30 .Male poor

ta | 20 : I > | | i ^ i1 O~~~~~~~~~~C]Femalenonpoor I ii..0 oo 18-25 26-30 31-35 36-40 41-45 46-50 51-55 56+ Agegroup

Source: World Bank staff calculationsbased on HBS-92.

1.19 The third important point is that education indicators - either enrollment ratios or literacy rates - vary greatly among governorates. Enrollment rates (as measured by the proportion of children between 6 - 18 years who are students) vary from less than 30 percent in Hajjah and Marib to above 50 percent in Sana'a City and Aden. (See Annexes II.C.) Adult literacy is highest in the governorates of Aden, Abyan, Sana'a, and Hadramout - at least 60 percent of the adult population is literate. It is lowest in Hajjah, Hodeidah, and Dhamar where it is less than 30 percent. This difference in literacy rates between southern and northern regions may be indication of the difference in emphasis put on education by the previous govemments.

AdultIlliteracyte. °l 50 - 0 . ,\ _ =~~~~~ 70

Al WtA < W

Source: World Bank staff calculations based on HBS-92

The map shows the illiteracy rate arnong the adult population by governorate. The denominator is the governorate's adult population -literate and illiterate. -8 -

Figure 1.3: Labor Force Participation Rat's by Sex and Poverly

Nonpoor Male -.... Poor Male

40 . ----- Nonpoor Female

. XPoor u Female

20

0 1S-20 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 Agegroup

Source: World Bank staff calculationsbased on HBS-92.

LABOR FORCE ACTIVITIES

PARTICIPATION

1.20 The labor force participation rates of the poor and nonpoor are 53 percent. The participation rate is defined as the share of the population (between the ages of 18 and 60 years) who are either employed or unemployed but seeking a job. The participation rate in Yemen is relatively low when compared to other developing countries primarily due to the low participation among women.

1.21 Figure 1.3 shows that the labor force participation rates by age groups differ among poor and nonpoor during their youth and when they are old. Poor men join the labor force sooner than nonpoor males who instead remain in school or college to obtain better educational qualifications. Poor men also have higher participation rates in later years (50 years and above) which most likely is a reflection of their need for income whereas nonpoor men can afford not to work.

1.22 Women's participation rate is low for both the poor and nonpoor at 18 percent. Poor women enter the labor force at an earlier age than nonpoor women. Their labor force participation rate declines during their early child-bearing years and then steadily increases as they grow older.

EMPLOYMENT STATUS

1.23 In 1992 the unemployment rate revealed in the 1992 household survey among 18 - 60 year olds was 7 percent of which slightly over one-third (38 percent) were first time job seekers.9 An estimated 9 percent of the poor are unemployed compared to 7 percent of the nonpoor. The unemployment rate among nonpoor young adults exceeds that of the poor which may indicate that the nonpoor can afford to search longer for jobs they feel are appropriate.

Accordingto Statistical Yearbook 1993, the rate of unemployment was 12 percentin 1991. - 9 -

1.24 The employed work mostly as self-employed and comprise 38 percent of the active labor force. About 80 percent of the employed poor work in the private sector either as self-employed or employees compared to 73 percent of the nonpoor. The public sector employs a significant share of the labor force (26 percent). In 1992 13 percent of all public sector employees fell below the poverty line as compared to 18 percent of private sector employees. This proportion has probably increased with the sharp decline in real wages of 69 percent during 1991 - 1994. (See Annex Il.D.)

SUMMARY HOUSEHOLD CHARACTERISTICS

1.25 Some characteristics of the poor and nonpoor are similar yet other aspects differ markedly. A selection of characteristics is presented in Table 1.4.

1.26 The characteristics which differ significantly by expenditure group are literacy of female spouses and male heads of household, average household size, and average number of under-18 year olds in the household. The difference in these characteristics between the poor and nonpoor exceed 20 percent with poor households possessing fewer literate male heads of household and wives, fewer female heads of household, and more children. The smallest difference between poor and nonpoor households is between the share of male children enrolled in school and share of children employed.

Table 1.4: Characteristics of Poor and Nonpoor Households

Ratio of Item Unit Poor Nonpoor All Yemen Poor to Nonpoor

Average household size no. of persons 8.5 6.9 7.2 1.23 Average number of under 18 year olds in household no. of persons 5.2 4.2 4.4 1.22 Average number of income earners in household no. of persons 1.7 1.6 1.6 1.05 Share of 6-16 year old children employed percent 2.8 3.1 3.0 0.97

Heads of household and spouses Share of literate male heads of households percent 44 57 55 0.77 Share of literate female heads of households percent 16 15 15 1.11 Share of literate female spouses percent 6 11 10 0.50 Share of female-headed households percent 3 4 4 0.87

Share of total expenditures on food percent 61 58 59 1.05 Share of total expenditures on qat and tobacco percent 10 12 12 0.81

hare of 6-18 year old children in school percent 40 41 41 0.99 Share of female 6-18 year old children in school p.:rcent 24 25 25 0.96 Share of male 6-18 year Aodchildren in school percent 54 55 55 0.98

Source: World Bank staff calculations based on HBS-92. - 10 -

GROUPS OF POOR

1.27 The Household Budget Survey does not capture important group characteristics and their impact on determining whether an individual is poor. Four groups with high levels of poverty are: 10

* Somali refugees of Yemeni descent who lost their assets during the civil war in Somalia. They are Yemeni citizens and have the right to work and live outside of UNHCR camps; an estimated 100,000 persons;

* Foreign refugees especially from Somalia who continue to live in camps administered by the UNHCR; estimated number is 10,000;

* Dispossessed farmers in southern Yemen due to the Government's decree that returned land to pre-Revolutionary owners without compensation; estimated at 100,000; and

* The Akhdam are an "ethnic group" of undocumented origin and occupy the lowest strata of northern Yemeni society. Upward mobility for persons in this group is difficult as is accessing services such as education for their children. Certain occupations such as sanitation-related ones (e.g., trash collectors) heavily employ the Akhdam.

AN OVERVIEW: REGRESSION ANALYSIS

1.28 A regression model is used to establish which characteristics are important in determining household expenditures. Results of the regression analysis are shown in Annex II.E.

1.29 According to the model, determinants of living standards as measured by household expenditures per capita are govemorate of residence, higher level education of head of household, and employment type. The variables which do not appear to make an impact upon expenditures are age and sex of household head and area (urban/rural) of residence. The model used here explains only 23 percent of the variation in expenditures which is average for such types of regression equations. Important variables, particularly on land ownership, agricultural production, and type of employment, are not included in the data set which could potentially have enriched our understanding of the causes of poverty.

1.30 The results indicate that expenditures per capita are strongly determined by where the individual lives. In particular if the household lives in Al-Mahwit or Dhamar, then its expenditures will be low compared to other governorates. Individuals living in Al-Jawf, Saadah, Al-Mahara, and Sana'a City will have a higher than average expenditures per capita. Households which reside in urban areas have higher expenditures than those in rural areas but only by a small amount.

1.31 The regression results show that the head of household's education (especially, post- secondary and primary school) matters in Yemen as it does in other countries. Though data are not available, this may indicate that the head of household's income increases with education. Post-graduate and university degrees have the greatest impact on expenditures. Additional analysis reveals that 97 percent of individuals with higher degrees are employed by the government.

'° Accordingto discussionswith NGOsand multilateralorganizations. - 11 -

1.32 The modeI reveals that households where the heads are either employers or managers of business or self-employed are more affluent tl dn others. Public sector employees are better off than counterparts in the private sector with comparable edi'cation which may explain the high demand for government jobs in Yemen.

1.33 The personal characteristics which impact expenditures per capita are household size and number of children. An interesting feature of the model is that if the spouse is literate, then expenditures per capita are higher.

MONITORING CHANGES IN POVERTY

1.34 The economic situation in Yemen has changed rapidly in the last few years since national unification in 1990. There have also been significant changes in consumer prices due to high inflation and changes in subsidies. The Government can obtain a clearer picture of how its economic policies are affecting the level and depth of poverty in the country by establishing a monitoring system.

1.35 A monitoring system would entail undertaking a small survey every one or two years with a sample frame based on the 1994 census. Prior to designing the survey questionnaire the Government should decide what the results of the survey would be used for in order to increase the relevance of the survey for public policy. Nationally representative surveys are generally expensive to undertake, and particularly so in i'.e case of Yemen due to its geography, low urbanization rate, and few rural roads.

1.36 The sample should be nationally representative but not as large as the 1992 Household Budget Survey which had 8,405 household observations. It is possible that surveying 3,500 households would be sufficient with 100 observations from each sector (urban and rural) in each of the 17 governorates plus Sana'a City. It would be more useful for analysis if the sample size from each governorate was of equal size rather than having a self-weighted sample. The results would be valid nationally, however, given the small size of the sample, drawing conclusions at the governorate level would most likely not be statistically robust.

1.37 In order to ensure that the poverty line adequately captures the minimum necessary resources required for living in Yemen, rural prices - at least for major food commodities - should be collected as well. In addition, the Government may also want to obtain further information on the food basket by analyzing food expenditure patterns in the HBS. - 12 -

2. SHORT-TERM IMPLICATIONS OF ECONOMIC POLICIES ON THE POOR

2.1 The macroeconomic situation has deteriorated over the last few years due to unfavorable economic policies, a hostile external environment, and a costly civil war. During 1990 - 1994 these factors caused slow economic growth (3 percent per annum), high inflation (57 percent), and an expanding budget deficit which reached 14 percent of GDP in 1994. The Government's inability to control public expenditures made stabilization and structural adjustment programs necessary to bring the economy into equilibrium.

2.2 These programs have unified the official and parallel market exchange rates and reduced the government deficit. They are expected to continue to lower the public deficit to sustainable levels through decreasing public expenditures (especially by reducing subsidies and transfers to public enterprises), and also to encourage private investment. The fear is that the poor will be hurt in the short-term unless steps are taken to implement programs which will provide direct assistance to them. (See Box 2.1.)

2.3 Though poverty is widespread in Yemen, it is primarily a rural phenomenon. Most of the population lives in the rural area. In 1992 the population was 14 million of which about 0.5 million were urban poor (4 percent) and 2.1 million (15 percent) rural poor.

2.4 This section finds that the Government's safety net costs nearly 16 percent of GDP and most expenditures do not assist the poor. The Government should consider reallocating these resources to other activities that benefit low-income groups, particularly social services. It is proposed that as a first step the poor can best be helped if the Government:

* Eliminates the wheat subsidy * Improves the import financing system to ensure continued supply of wheat after market liberalization * Supports the expansion and institutional development of the nongovernmental sector to deliver assistance to the poor

THE GOVERNMENT'SSOCIAL SAFETY NET: UNIVERSALSUBSIDIES

2.5 The Government has subsidized food, petroleum, water, and power to protect consumers from paying the full cost of these goods. Though some consumers have benefited from these subsidies since they pay lower prices, they have also been hurt because these expensive subsidies (estimated at 16 icrcent of GDP in 1994) have reduced growth and fueled inflation.2 Government subsidies and other untargeted public welfare programs benefit primarily urban dwellers where goods and services are provided. In order to understand the magnitude of resources consumed in subsidies, I percent of GDP could provide a basic health package to 3.7 million persons (oi 27 percent of the population).

The proportionof the rural populationis estimatedat 80 percent of total populationusing census data from 1986 and 1988. Updated informationfrom the 1994 censusis not yet available. 2 See Republic of Yemen: Dimensions of Economic Adjustment and Structural Reform for a more detailed discussionon electricity,petroleum products, cement, water, and waste water subsidies. - 13-

Box 2.1

STRUCTUrLADJUSTMENT AND THE POOR

A new study by the Bank dispels some common concerns about adjustment lending and poverty. The study looks at all the evaluated adjustment operations that the World Bank supported in 1980 - 1993 - 144 operations in 53 countries - and tracks what happened to poverty and income distribution. The results show that countries that successfully implementedthe adjustment policies agreed with the Bank have achieved growth in per capita income and reduced the proportion of their populations living in poverty. The study stresses the need for the Bank to improve the success rate of its adjustment operations, to promote reformnsthat bring the poor into the growth process, and to improve the quality of public expenditure, so as to safeguard vulnerable people.

The study shows that adjustment lending is a valuable instrument of reform: two thirds of the 53 borrower countries implemented the adjustment programs with a substantial measure of success. They reduced price distortions and inflation, improved their resource balances, stabilized their foreign exchange reserves, and achieved growth in per capita income. They reduced their restrictions on trade and capital flows, integrated themselves more fully into the world economy, and became better able to use their resources efficiently.

For 23 countries that borrowed for adjustment, enough poverty data are available for reliable "before and after" comparisons. Two thirds of these countries successfully implemented adjustment, achieved per capita income growth, and all but one (Kenya) reduced the percentage of their populations living in poverty. The findings affirm that well-managed adjustment can help remedy social distress. This said, rates of poverty reduction varied widely. Over the adjustment period, reductions in the proportion of poverty ranged from more than 3 percent a year in Indonesia and Thailand to 0.2 percent in Chile and Senegal. Poverty declined in all countries studied in Asia, but only in half of those studied in Latin America. Countries that delayed or postponed adjustment, such as Argentina, Brazil, Peru, and Venezuela, saw poverty worsen further in the 1980s. In Africa, Ghana and rural Tanzania displayed clear signs of economic recovery and poverty reduction, while in rural Kenya and Cote d'lvoire growth was sluggish and poverty increased.

Source: OED Precis (October 1995) which reviews the report, The Social Impact of Adjustment Operations bv Carl Jayarajah, et al.

2.6 A decrease in direct and indirect subsidies would have a negative impact on the welfare of some groups, especially urban residents. In those cases where they consume the subsidized goods, elimination of subsides would lead to price rises. Two poverty-related effects may result. First, the depth of poverty may increase because the urban poor would suffer a decrease in their purchasing power. Second, the incidence of poverty may increase because some households which were just above the poverty line will fall below it as their purchasing power erodes from the price increases on previously subsidized goods.

2.7 The rural poor mostly depend upon subsistence agriculture to meet their basic needs. In those cases, where the rural poor enter the formal market for either supplying or purchasing goods and services, they could be affected negatively or positively by the decrease in subsidies. This depends upon whether they are net consumers or net producers of the subsidized goods. Particular subsidies such as the wheat subsidy may have actually hurt the rural poor, while the petroleum subsidy may have helped them by keeping the cost of transportation low. - 14 -

Figure 2.1: Wheat Imports and Subsidy

7 6

u 4 ~ _,Subsidy 05C 3 Imiports V) 2~

0 1992 1993 1994 1995 1996 Year

Source: Ministry of lFinanceand Central Banikof Y emen.

2.8 In the short-term the removal of subsidies will increase prices, thereby decreasing the purchasing power of the urbanipoor and nonpoor. However, the continuation of these subsidies is one of the primary sources leading to a high inflationiaryenvironment because the government borrows or prints money to maintain them. Since the poor usually lack assets and cannot hedge against inflation, general price appreciation (even with subsidies) hurts them more than the nonpoor who possess land, houses, and vehicles whose prices keep pace with inflation. Thus, the net effect of subsidy reduction on the urban poor is uncertaini.

WHEAT SUBSIDY3

2.9 Subsidies on rice and sugar were .iminated by Tnid-1994 and the wheat subsidy is the last remaining food subsidy. It is projected to accoun!:for 4 pei-cert of GDP in 1995 which is 25 percent lower than its share in the previous year.

2.10 Several major issues with regard to the wheat subsidy are discussed below. Though all information indicates that the wheat subsidy should be eliminated (the Government itself has decided to reduce it), there are certain considerations which should be taken into account in choosing the timetable and mechanism.

MAJOR ISSUES

2.11 The wheat subsidy is expensive. In 1994 wheat imports totaled $273 million and are projected to rise to $400 million in 1996 or 8 percent of GDP. The high cost of wheat imports has contributed to a large current account deficit and the inability of the government to pay for wheat on a timely basis, which has added significantly to the cost of imports. As of mid 1995, the Central Bank was in arrears for past wheat imports by $160 million. (See Figure 2.1.)

3 The wheat subsidyrefers to both wheatand flour subsidies. - 15 -

2.12 The wheat subsidy benefits mostly the nonpoor. In 1995 only a maximum of $65 million of the $254 million subsidy is estimatedto have reachedthe poorest 20 percent of the population,though it is likely to be far less than this amount.4 There are three reasonswhy the wheat subsidy may not help many of the poor. First, though the Governmenthas established wholesaleand retail prices for wheatand has distribution centers,much of the wheat is diverted from the governmentdistribution channelsinto the private market. Also, the re-export of an estimated20 percentof wheat imports to neighboringcountries results in pushing the domestic price even higher. This is a governmentsubsidy of $51 millionl (1 percentof GDP) that benefits foreignersand re-exporters. Once the wheat is in the private market, the prices rise above the control retail price of YR 277 per 50 kilogram to the equilibriuimmarket price of YR 500.

2.13 Second,since wheat is consumedmostly by the urban population, the majority of the poor who live in the rural sector do not benefit from the subsidy program. They consumemore traditional cerealssuch as sorghumand barley.

2.14 Third, the urban poor benefit from the subsidy mainlly by consumingcheaper bread rather than purchasingwheat or flour. This is becauseurbani distribution centersare few and rarely locatedin poor urbanareas- government stores only sell wheat in 50 kg sackswhich often exceed the purchaslingcapacity of poor families; and constumiiptionof bread at home would requireadditional expenditures on fuel.

2.15 Wheat consumersbenefit from the subsidyprograin whille domesticgrain producersare hurt. Despitethe high cost to the country, consumersdo benlefitfrom the wheat import subsidy becausethe system oversuppliesthe domesticmarket withi wheat - perhapsin an attempt to compensatefor the leakagesout of the governmentdistribution chaniels. This lowers domestic prices to all consumersbut not to the level specified bv the Government. For example, the average import price of wheat flour in 1994 was $223/MT compared to the actual consumer retail price of $84 and the (unenforced)Government control price of $47.5 If we assumed (perhapsunrealistically) that wheat was both price and income inelastic and that the rural poor purchasedwheat, then a 50 percent rise in the price of xvheatwoould increase poverty from 22 percentin 1995to 24 percent.6

2.16 It is difficult with the existing datato quantify the impact the wheat subsidy has had in suppressingincomes of grain farmers, and thus on the affecting level of poverty in the rural areas. In aggregate terms, however, an additional consequenceof the subsidy is the discrimination against domestic wheat producers implying an estimated$16 million transfer from the domestic producersto consumerscompared with the free market import case. The systemalso isolatesdomestic producers and consumersfrom most of the changesin international prices. Both the supply responseof producersand the demand responseof consumersare reduced,resulting in greater imports than would be the caseif interinationalprice signals were fully transmittedto the domesticmarket.

Accordingto preliminaryresults of the Public Expenditure Review (World Bank 1996),only 3 percent(or US$ I I million) of thetotal wheat subsidy is capturedby thepoorest 20 percent. The dataare for September1995. The sourceof informationon the averageimport price andthe Governmentccntrol price is the Ministry of Supplyand Trade. The averageretail price was estimatedby missionmembers based on a small nonrepresentativesample. The parallelmarket exchangerate of YR i I 9/US$was applied to calculatethe pricesin US$.

6 This is det,rrninedby recalculatingthe governoratepcvertv lines using wheatprices that are 50 percentabove tne actual prices given in th-eSittistical Yearbook, 1993. - 16 -

2.17 Wheat is a major component in total food supply. The removal of subsidies and the freeing of imports on other food items (such as sugar and rice) has not appeared to cause major problems for the majority of the population. However, wheat is a primary staple, especially in urban areas, and it may not be as easy for the Government to end the subsidy. Average wheat supply per capita is 154 kg per capita per year in Yemen compared to 143 kg in Jordan. Wheat accounts for about two-thirds of total cereal supply (imports and domestic production), though not all may be consumed domestically due to illegal exports. The supply of wheat grew by 9 percent per annum from 1980 - 1994 aided by large imports and subsidized consumer prices. Currently 94 percent of wheat supplied is imported.

2.18 The Government of Yemen pays more for wheat than the international market price. Wheat imports cost an estimated 19 percent more than world market prices. In 1994 this meant that the Government paid $47 million dollars more than it could have because of poor purchasing, financing, and handling practices. These charges result largely from risk premiums which foreign suppliers and domestic importers add to their offer prices to cover delays in payment, penalties for slow unloading, unauthorized port charges, and a failure of the Government to abide by contract terms on wheat imports. The domestic distribution system is government controlled and results in overcharges and rent seeking. Monopoly power is granted to a government corporation for domestic distribution in each of four regions and private agents are granted monopoly power to sell government wheat and flour to the consumer.

OPTIONS

2.19 The wheat subsidy can be eliminated by increasing the exchange rate for wheat imports to the parallel market rate and by liberalizing the import and distribution system. The wheat subsidy is enacted through the exchange rate: for example, in 1994 the Government spent $303 million on wheat imports and sold it to wholesalers for US$ 45 million. The additional benefits of removing the wheat subsidy are that smuggling across intemational borders will be reduced, thereby ending the Government's subsidy of foreign nationals in neighboring countries.

2.20 Price liberalization of wheat should also be accompanied by a liberalization of the distribution system. The current system grants distributors (wholesalers and retailers) a defacto regional and local monopoly respectively which results in consumers paying a higher price than they would if the procurement and distribution system were competitive. The Government should (i) abandon the current arrangement which gives retail agents sole right of distribution in a particular locale and (ii) allow importers to decide who they will sell the wheat to, thereby removing the public sector monopoly on wholesale distribution as well as governorate wheat quotas. Yemen has a well-developed grain market which successfully distributes other grains such as sorghum, rice, and barley.

2.21 The government outlay for the wheat subsidy program can be reduced by an estimated 30 percent without significantly hurting the consumer if a phased approach to subsidy removal is followed. This requires Govemment commitment to (i) reducing the margins on transportation and handling by renegotiating prices with the trucking union and port handlers; (ii) complying with contracts made with exporters and importers; and (iii) providing secured letters of credit for import commitments. These last two actions would decrease the risk premium added to the international market price because of the Government's poor reputation in honoring contracts. In addition, if the Government improved the monitoring of the distribution system to ensure that the savings reach the consumers, additional savings could be made. - 17-

2.22 The Government can re-establish its reputation in the international wheat market which would lower the purchase price of wheat - a benefit that the Government could pass onto consumers. In order to accomplish this, the Government of Yemen may want to consider (i) clearing its arrears of $160 million over the next 3 to 6 months, perhaps by setting aside some of its oil revenues; (ii) establishing a "wheat fund" where foreign exchange reserves are set aside at the time the Government finalizes the contract with importers; and (iii) reaching bilateral agreements with neighboring countries to discourage illegal trade in wheat and flour. If these steps are taken, the Government can avoid paying a risk premium for wheat imports which could result in significant savings. For example, in 1994 the risk premium is estimated at US$ 47 million or I percerit of GDP.

SOCIAL SAFETY.' NEIS: PUBLIC OR PRIVATE INITIATIVE?

2.23 Universal subsidies are expensive programs since leakages to the nonpoor are large and only a small share of the benefits reach the poor. One immediate response to poverty alleviation is to have the Government redistribute income from the nonpoor to the poor through cash or in- kind transfers, that is, establish a social safety net.

2.24 A formal safety net needs to deliver a small amount of assistance to a large number of people thereby raising the poor's expenditures closer to the poverty line. A well-functioning program minimizes leakages to the nonpoor. Establishment of safety nets have been successful mostly in OECD countries, transition economies, and some developing countries (such as Jordan) which possess better administrative systems for targeting and delivering assistance to the poor.

MAJOR ISSUES

2.25 Direct eradication of poverty would cost Yemen at least 5 percent of GDP. One possible way of eliminating poverty is to provide cash or in-kind transfers to the poor sufficient to raise their expenditures to the poverty line. In the case of Yemen, the average transfer to the 3.4 million poor in 1995 would need to be $33 per capita per year.8 In addition to transfers, other expenditures would include administrative costs and a certain amount of leakage to the nonpoor. The total cost of such a program would be about $i59 milliou per year.9

The risk premium on wheat imports is estimatedto be the difference between the cost of Yemeni imports and the cost available from the world market. For 1994, these figures were $166/tonand $140/tonrespectively. Total importswere 1.8 milliontons. 8 This calculationis basedon the 1995 exchangerate for GDP conversion. 9 This calculationassumes that (i) administrativecosts and leakagesto the nonpoor would be about 20 percentof transfersand (ii) childrenand adultswould receiveequal transfersper capita. - 18 -

2.26 Yemen does have a transfer program targeted to the unemployable poor. According to the Ministry of Social Affairs, Social Insurance, and Labor (MOSL), about 31,000 households are assisted with cash and food assistance estimated at YR 31 million or US$ 387,000 in 1994). Average benefits per capita are YR 200 (US$ 15) per capita per annum.10 However, significant leakage to the nonpoor is reported due to the Ministry's difficulty in verifying the benefit claims of applicants.

2.27 Yemen has strong nongovernmental support systems which provide targeted small-scale assistance to the needy. There are three types of informal support systems in Yemen: (i) religious charity (zakat and satqa) , (ii) traditional community-based assistance (such as communal voluntary work, assistance for weddings, labor pooling among neighbors, and education support), and (iii) within-family transfers. Althoughi it is difficult to estimate the amount of assistance or the extent of coverage provided by these activities, there is sociological evidence that these systems function well. This is why, according to some scholars,12 starvation due to lack of financial support rarely occurs in Yemen. Some believe that demographic and socio-economic pressures have started to erode some of these systems. Systematic sociological and economic research would help us to understand better the strengths and limitations of nongovernmental assistance.

2.28 Targeting cash or goods to the poor in Yemen is difficult. A targeted safety net needs to distinguish between the poor and the nonpoor. This is particularly difficult in the case of Yemen because most of the poor live in rural areas in small dispersed communities. Furthermore, since household expenditures' at lower levels do not vary much as seen in Annex 111,it is difficult to distinguish between those people just below the poverty line and those just above it. Though three types of targeting mechanisms are available (means-testing, self-targeting, and geographic targeting), some have less potential than others.

2.29 Means-testing determines the level of a household's (or individual's) income and wealth or "means" in order to decide whether the household is poor and qualifies for assistance. This targeting mechanism would be the most difficult to implement in Yemen. The prevalence of informal economic activity in urban and rural areas as well as intra-family transfers makes it difficult for the Government - even if it possesses the necessary institutional capacity - to measure a household's income and verify the respondent's answers.

2.30 Self-targeting mechanisms are those which discourage the nonpoor from participating in a transfer program. There are basically three types of mechanisms: time, stigma, and type or quality of a product. Safety net programs which use self-targeting include those which require recipients to stand in line for lengthy periods of time to collect a commodity or cash; subsidize inferior quality food or other commodities consumed by the poor rather than the nonpoor; and public works program. In Yemen insufficient information is available on the food baskets of the

The total cost of the cash and food aid programare YR 25 million and YR 6.1 million respectively. The food aid programprovides 12 kg wheat flour, 600 grams each of cooking oil, beans, and sugar per capita up to a maximumof five personsper household.

l Zakat is 2 to 5 percentof wealthwhich is given to the poor. Satqa is voluntarycontributions of food, money, and clothesdonated to the poor. 12 Discussionswith the faculty of the Departmentof Sociology,University of Sana'a. - 19 - poor and nonpoor for determining whether self-targeting of goods is an option.13 Public works programs use either the wage level or the type of work as a means of targeting assistance to the poor. This can be successful if the target group is unemployed or underemployed. (In 1992 only 9 percent of the poor are unemployed.)'4 If the work is sufficiently unpleasant or wages sufficiently low, then the nonpoor will not have an incentive to participate in the program.

2.31 Geographical targeting would necessitate the provision of goods which are large components of the poor's consumption basket (e.g., wheat, sugar, cooking oil) at subsidized prices to locations where the nonpoor would not go, such as stores in slum areas. The difficulty with this option is that opportunities for individuals, including the nonpoor, to make profits will be high which will result in diverting benefits from the poor to the nonpoor. However, if exploitation of the system could be controlled, provision of subsidized good in governorates with high concentrations of poverty such as Dhamar, Al-Beida, and Al-Mahwit is a possible option to explore.

Table 2.1: Local Government Contributions to Social Welfare

Sector Activity LCCD 1985/86 Transport Roadconstruction 852km Road maintenance 3,836 km Education Classroomconstruction 783 classes Schoolmaintenance 1,125classes Water Potablewater projects 267 projects Potablewvater maintenance 66 projects Health Clinic construction 71 rooms Clinic maintenance 148 rooms Other Parks,mosques 261 projects Maintenance 141 projects

Source: Democratic Developments in Yemen, (ed.) Amat Al-Alim As-Suswa.

2.32 Yemen's community development initiatives have diminished in importance because of the interference of the central government. Both the former YAR and PDRY had established local community development organizations. In north Yemen they were initially called Local Development Associations (LDA) and, more recently, Local Councils for Cooperative Development (LCCD). These organizations were community based and used local contributions for projects such as roads, clinics, schools, and parks. In the first phase of establishment, they were autonomous and worked with the central government rather than being under its direct control. In south Yemen, People's Local Councils and the Popular Defense Committees undertook similar activities; however, from the beginning they were controlled by the central Government without power to act independently. (See Table 2.1 for an overview of projects.)

3 A nutritionalsurvey is required to determinewhether the poor consumegoods which the nonpoor do not. If such items exist, then these goods could be moderately subsidized without incurring significantleakage.

X Accordingto some offical sources,the unemploymentrate is between 30 and 35 percent. However, since there has been no recent survey,this estimateis most likelybased on an unrepresentativesample from urbanareas ratherthan reflectingthe nationalsituation. - 20 -

2.33 The Government's increased involvement in the local organizations essentially made them extensions of the central authority rather than local governing bodies and local residents lost their commitment as decision-making was overtaken by Government representatives. As a consequence voluntary local contributions of time, labor, and money - as well as development projects - have fallen.

2.34 The domestic nongovernmental sector is small and lacks institutional capacity and access to financing. Other forms of community development which are rapidly evolving in Yemen are NGOs and professional cooperatives. Though many are registered with the Government, few are actually operational or undertake welfare activities. About fifteen NGOs exist which are actively involved in delivering reasonable quality health, education, and other social services to the population mostly in urban areas. These NGOs typically are successful in targeting assistance to the lower income groups. However, most NGOs have limited capacity for planning or implementing projects as well as few resources to upgrade themselves or deliver benefits to a large rnomberof poor.

2.35 The public sector has supported the expansion of the NGO activities. Selected NGOs and cooperatives receive some financial support from MOSL. For example about twenty NGOs and cooperatives aeceived YR 20 million from the Government in 1994. The criteria for support, however, are iiot clear and are based on the Ministry's judgment of the NGO's size, effectiveness, and importance. In addition to the Government, NGOs raise funds from individuals and businesses. Bilateral international assistance to NGOs is still limited due to the absorptive capacity of these organizations.

OPTIONS

2.36 The Government should return development councils to community control. The most important lesson from the LDA and LCCD periods is that government control discourages local initiatives. Though it is debatable whether LCCDs can be revived, the spirit of cooperation for community development is still present, and the Government should not interfere in these initiatives. The very essence of their enterprise lies in their spontaneity and local relevance. The Government should encourage the expansion of community development organizations, including NGOs, by providing assistance such as counterpart funds or technical advice when it is sought.

2.37 The Government should rationalize its criteria for giving financial support to NGOs. Since the Government is an important source of financing for the nongovernmental sector, it is essential that it remains impartial. In order to support effective NGOs, the Government should establish formal, results-based criteria for public assistance. For example, to improve living conditions for the rural poor, it could give greater financial support to NGOs working effectively in remote or rural areas. Greater transparen-y ir, unding m'ay lead to better functioning NGOs and better living conditions for the poor.

2.38 The NGO sector needs assistance to expand provision of welfare and service delivery p2rograms. NGOs can provide alternative and cost effective methods of reaching the poor if they are equipped with the necessary managerial and technical tools. In order to promote the efficient delivery of good quality services to the poor, the Government should develop a comprehensive policy conducive to the expansion of NGOs. The policy would include incentives for helping the poor, seeking ways to improve intemational donor funding coordination, training, and supervision and monitoring of projects funded directly by the Government. The proposed Social - 21 -

Fund Project supported by the World Bank will assist in establishing a mechanism of encouraging the development of and financial support to the nongovernmental sector, microenterprise activity, and social services delivery to communities.

2.39 The Government needs to develop instruments to help the poor. At present, it is difficult for the Government to deliver aid to the poor at a reasonable cost and without significant leakage to the nonpoor. However, it should begin to develop a plan for implementing a targeted safety net. Economic growth will raise many people out of poverty but may not help those who cannot work such as the disabled and the elderly.

2.40 As a first step, the Government can improve the Cash and Food Aid program targeted to the unemployable poor by seeking ways to decrease the reported leakage to the nonpoor. Second, it should expand the program to cover smaller urban centers. Third, the Government should determine mechanisms for targeting the unemployable poor in rural areas and delivering cash transfers to this group. Design of such programs should be undertaken with input, and preferably participation, of targeted beneficiaries. In addition, MOSL can work with the Department of Statistics to obtain an estimate of the target group's size; it can also use infonnation provided by the beneficiary assessment (which is being undertaken as part of the preparation of the World Bank Social Fund project) on coping mechanisms in Yemen. Fourth, the Ministry of Health should obtain estimates of the level of under-nutrition in Yemen and work with MOSL to determine to what extent this is due to poverty or poor dietary habits (e.g., weaning practices for infants). - 22 -

3. MEDIUM-TERM IMPLICATIONS OF ECONOMIC POLICIES ON THE POOR

3.1 Poverty in the medium-term can be rmduced through redistribution of income and economic growth. Government intervention in the economy can facilitate these goals through regulations supporting economic competition, provision of public goods and services, or assisting the poor directly. As discussed in the previous section, since redistribution of wealth from the nonpoor to the poor has limited potential in Yemen, due to the difficulty in targeting the poor, the country will have to rely on economic growth to reduce poverty.

3.2 Economic growth increases a society's total wealth. The expectation is that as the country's wealth grows, the poor will benefit proportionately at least. However, much of Yemen's recent economic growtlh has resulted from petroleum and natural gas exports. The oil sector is large-scale, capital-intenisive, and employs a relatively small share of the labor force. The agriculture sector though smaller in terms of its contribution to GDP as compared to oil, provides employment and income to a far greater shiare of the population. Thus, the type and sources of economic growth matter for poverty reduction.

3.3 Oil revenues accrue to the Government and allow it to reduce taxes or increase the activities which the public sector can finance. The (Government must decide whether the oil income will fund public investment, public consumption, or lower taxing and/or borrowing. The type of public investment the Government chooses to fund has different implications for its recurrent and capital budgets.

3.4 Poverty alleviation in the medium-term requires a combination of economic policies fostering broad based economic growth and the use of public resources to improve the welfare of lower income groups. This chapter proposes that the level of poverty can be reduced if the Government takes the following steps:

- Reevaluates the allocation of governmnentexpenditures * Rationalizes public investmlenitto reflect sectoral and regional priorities * Rationalizes agriculture sector policies

THE FOUNDATIONSFOR ECONOMICDEVELOPMENT

3.5 Yemen's population is growing at 3.3 percent per year. In order for Yemen to achieve even a modest improvement in the standard of living, the economy must grow by at least 4 - 5 percent per annum. The Governm1entcan best achieve this by: (i) macroeconomic stabilization, (ii) structural reform, and (iii) social protection during adjustment. In order to achieve macroeconomic stabilization- a prerequisite for economic growth - the Government must decrease its fiscal and trade deficits to sustainable levels. Structural reform requires trade policy reform, privatization of public enterprises, and regulatory reform to improve the investment environment. Social measures during adjustment include protecting public expenditures on health and education. - 23 -

3.6 An outward-oriented economy is essential for Yemen to enjoy the benefits of international trade. This can best be achieved through development of the tradeable goods sectors, that is, industry and agriculture. Currently, these two sectors contribute 55 percent to GDP of which oil-related activities contribute 22 percent to national output. Yemen's petroleum and natural gas industries have expanded rapidly in the last few years, however, this has not been accompanied by improvements in productivity or wages for a significant share of the labor force due to the capital-intensive nature of the production process. For the population to participate in economic growth. greater investments in export-oriented labor-intensive (manufacturing) activities need to be made which will lead to an improvement in wages.

3.7 Though it is not possible to predict the future sources of economic growth for a country, analysis can reveal which areas have been overlooked in the development process. In the case of Yemen, for example, the agriculture sector - despite its many constraints - could be more profitable if barriers to trade where removed. The manufacturing sector may have strong potential even in the short-term if regulatory obstacles were reduced (such as the licensing and bureaucratic approval procedures). Growth in natural gas exports could help to finance the development of other sectors as well as human capital improvements - which is essential for raising output in the medium- to long-term. Other potential areas which have been under- exploited are fisheries, tourism, and manufactures exports such as handicrafts.

3.8 In addition to macroeconomic stability, three other prerequisites for sustained growth are the reform of the financial sector, legal code, and civil service. Broad based economic development depea Is heavily upon private investment which requires a functioning financial sector to raise mapita,as well as a stable legal environment where contracts and property rights can be enforced. The public sector also plays an important role in the economy, especially in Yemen. Current government hiring practices have adversely affected the efficient delivery of public goods such as education and health services. In many iow income countries, the quality of public service has suffered as its ethos has jeen destroyed by a combination of over-staffing and inadequate pay. In Yemen the situation is of particula! concern, because of the large role the Government plays in the formal economy and in the deliNery of essential goods and services.

3.9 Since Government action can have important impacts on growth, the sections below discusses public expenditures. The budget is analyzed from various perspectives including governorate-level, current and capital categories, and wage and non-wage operating costs.

MAJOR ISSUES

3.10 The Government plays too large a role in the economy which inhibits growth. The share of government expenditures in GDP is about 38 percent during 1990 - 1995 - significantly higher than the average for low income countries (24 percent). The public sector also employs one out of every four labor force participants. The over-involvement of Government in a country's economic activities can have a detrimental impact upon the country's long-term development as seen in former socialist economies such as the Soviet Union and the PDRY. The more resources a government absorbs, the fewer remain for private investment or consumption. Furthermore, despite the heavy involvement of the Government in Yemen's economy, there is inadequate provision of essential goods and services (e.g., health, education, and sanitation) which improve living conditions for the populatior.. - 24 -

3.11 The Government is living beyond its means and has not controlled public consumption. Government revenues covered on average 57 percent of total expenditures during 1990 - 1994. In 1995 the Government made significant progress in managing the country's finances: 73 percent of public expenditures are now paid for by current revenues compared to 45 percent in 1994. However, governments like households cannot live beyond their means for extended periods without serious difficulties. If a country is to avoid economic disaster, consumption and spending must balance with its income or revenues in light of its growth.

3.12 The correlation between government revenues and current expenditures is small which implies that irrespective of its income, the Government perceives it has little flexibility in decreasing current expenditures. For example, during 1990 - 1994 total nominal expenditures increased 29 percent annually while revenues increased by 7 percent only. As a result, inflation accelerated in Yemen and reached 72 percent in 1994. The Government tried to mitigate the effects of inflationi on the urban population by increasing consumer subsidies, especially on wheat, which only served to heighten the level of fiscal instability rather than help the poor.

Figure 3.1: How the Government distributes the recurrent budget

50 -

,' 45 -- 40-

C. 35 c30______rs___ __| | i Wagesy i ED& Salaries DefenseExpenditures C]r Goods&Services CurrentTransfers

0Q 10I

0 5 0 1990 1991 1992 1993 1994 1995

Sources: World Bank staff calculationsbased on Ministrv of Financedata.

3.13 The composition of Yemen's recurrent budget is skewed away from expenditure categories which can foster economic growth. As seen in Figure 3.1, only half of the Government's recurrent budget is allocated to the fundamental activities of administering a country - wages and salaries and goods and services. Defense expenditures and current transfers absorb the remaining budget. Yemen's defense expenditures as a share of public expenditures are high (34 percent) when compared to other countries in the region but relatively modest in per capita expenditure terms ($35). (See Annex IV.A.)

3.14 The central government budget itself is not well allocated. Wages and salaries at 81 percent are a disproportionately high share of government recurrent expenditures (excluding defense and current transfers). For employees to be productive they need access to sufficient quantities of materials, such as medicines for doctors, textbooks for teachers, and stationary for administrators. As the productivity of government employees fall, so does the effectiveness of services provided, especially in such sectors as health and infrastructure. - 25 -

Table3.1: Existing Infrastructurein the DevelopingCountries Power Generating Paved Capacity Telecommunications Sanitation Rod, O00sofkws per mainlines per 100 %of %of pop km per mi; ml persons persons population with access persons with access to safe water Yemen 21 1 30 64 181 Low income countries 53 6 42 62 396 Middle income countries 373 8 68 74 1,335

Middle East countries Algeria - 32 - 77 1,366 Egypt 230 4 45 88 267 Jordan 269 7 1 0 99 i,885 Lebanon - 9 - - 1,224 Morocco - 16 - 56 618 Syria 286 4 55 93 1,885 Tunisia - 38 - 70 1,177 West Bank/Gaza 2.2 - 90 1,177 S3urce: "Infrastructure Investment in the Middle East". Inder Sud.

3.15 Public investment has decreased sharply since unification. The Government reduced the fiscal deficit by decreasing public investment from 29 percent in 1990 to 8 percent in 1994. However, since its share in total expenditures is quite low, continuing to reduce it will have minimal impact on the deficit. The capital budget has suffered the most from decreasing current revenues. Though investment has declined significantly since 1990, the decline is less than what the budgetary figures indicate because the overvalued exchange rate underestimates the contribution of imports in public investment.

3.16 Between 1990 and 1994 (adjusted) public investment fell from a high of 14 percent of GDP to 6 percent which is about average for low-income countries. However, once public investment is disaggregated into core "development" activities (such as construction of roads and schools) and capital transfers, the share of development investment in GDP falls to 2 percent in 1995 which is unacceptably low. Table 3.1 highlights the relative lack of almost all categories of infrastructure in Yemen even when compared to other low income countries.

3.17 Unless development of the rural areas occurs. poverty will remain high in Yemen. Most of the poor live in the rural areas and depend upon agriculture either directly or indirectly for their survival. Since development of the non-oil industrial sector is projected to be 5 percent per annum during 1995 - 2000 from a base of 14 percent of GDP, the absorptive capacity of this sector is limited. Thus, for economic growth to occur and for a larger share of the population to participate in it, the other remaining tradeable goods sector - agriculture - must play a stronger role. Though Yemen faces a serious water constraint, the issue becomes one of using water efficiently and in high payoff activities.

3.18 Though data is limited, poverty appears to be concentrated in governorates with the highest concentration of small farms (Dhamar, Al-Mahwit, Taiz, lbb, and Saadah). Thus, unless small farms can be made more profitable by increasing the value of the yield (either by switching to high value crops or by increasing the yield per hectare) or decreasing the cost of production and marketing (such as through improving infrastructure), the poor will remain poor. - 26 -

Figure 3.2: How Economic Growth Affecds Povefy

2015 => _ I-- '' io~~~~~~~~~~7%l a Q 2010

=._ . l t \ .~~~~~~~~~~~3% o. P 2005 E iI

cjn .2 2000

0 10 20 30 40 50 60 Year

Source: World Bank staff calculations based on HBS-92.

3.19 Government policies discriminate against the agriculture sector which employs the largest share of the population as well as the greatest number of poor. In particular the official exchange rate effectively subsidizes imported wheat and flour making it difficult for domestic farmers to compete. As a result domestic grain production has been stagnant or decreasing. Farmers who have been able to switch to high value crops, such as qat or fruits, have mitigated the impact of government policies on their income. Those farmers that cannot switch to more profitable crops are forced to bear the consequences of these policies.

3.20 The pervasive use of qat in Yemeni society partially redistributes oil revenues from the urban residents to rural residents and from tradeables to nontradeables. The purchase of qat by male adults diverts 12 percent of household expenditures towards the agriculture sector (estimated at $188 million annually for the urban population') which may explain the relatively small per capita expenditure differential between urban and rural residents in Yemen.

OPTIONS

3.21 In order to reduce poverty by at least fifty percent within the next 10 years. Yemen's economy must grow by 6 percent per annum. If the rate of economic growth (as measured by GDP) is 3 percent or less, poverty will increase over time as more people share the same amount of wealth. Without any growth, the number of poor will increase to over one-third of the population by the end of the century due to the high population growth rate estimated at 3.3 percent. Figure 3.2 illustrates changes in poverty if we assume the distribution of expenditures remains constant with economic growth (i.e., a poverty to growth elasticity equal to one). If inequality decreases, growth will accelerate the rate of poverty reduction while a deterioration in the distribution will mean that an even higher rate of growth will be needed to have the same impact on poverty.

This calculationis based on HBS-92. It is somewhat higher than a more "back of the envelope estimate"of $1 18 ml where 60 percent of the adult male population(about 25 percent of the total populationis assumedto purchaseYR 100 worth of qat per day for 330 days per year. The estimated expenditurein 1995 is $592 million at the parallelmarket exchangerate for all Yemen. - 27 -

3.22 The Governmentshould reallocate its expendituresin order to improvethe environment for economic growth and lay strong foundations for long-term development. The current allocation of resourcesbetween recurrent and capital expenditures and between different recurrentcategories may inhibit the effectivenessof Governmentintervention in the economn. The Governmentmay want to considerthe following steps:

* First, the Governmentshould reevaluatethe cuts in materialexpenditures (Chapter 2 of the budget) which have occurred over the last few years. This category has been downsized during years of high fiscal deficits. Without adequatematerials, public sectoremployees do not have the meansto fuilfill their responsibilities.

* Second,the Government should reconsiderthe high pace of public sector hiring. Durinig 1990 - 1994, the number of public sector employeeshas increasedby about 50 percent though the shareof wages and salaries in GDP has remained constantat 23 percent. As a result, there has been major depreciation(of about 69 percent in real terms) in the average real wage of public sector employeesduring these last four years: poverty amonigcivil servantsis projectedto have increasedto 24 percentIn 1995. If real wagescontiniue to fall, a greater shareof employeeswill fall below the poverty liie which will put pressureon the Governimentto increasesalaries.2 (See Aniex IV.)

. Third. the Government should aim to increase domestic investment in "development- expenditures. Without investmenitthe population's living standardsare determined by current income. If Yemen wishesto achievea future level of developmenitand consumption which are higher than what it possessestoday, the Governmrenitneeds to encouragethe expansionlof the country's existing stock of capital, both humaniand physical.

Developmentexpenditures should be twice the existing capital budget. In order to reachthis level over the next few years without iicreasing total governilmelitspendilig, significant clhangeswill needto be madein the way the Governmzentcurrently allocatesits budget.

3.23 Measuresshould be taken to improve the profitability of the agricultule sector in the medium-term. Yemen is primarily a rural couLntrywhiere the agriculture sector provides employment and income to a significant shareof the populationi(about 58 percent of the labor force), particularly low income groups. Though)the agriculture sector is unlikely to absorb a greater share of the labor force in the fuiture, it would be an oversight to dismiss its role in economicdevelopment of Yemen. Furthermore,since yields are relatively low and much of the production (includinig qat) is for local markets, possibilities exist to increase its export orientation.

3.24 Grow-thof aggregateagricultural output typically takesseveral years to materializesince it requirestechnical changeand net additions to the stock of labor and capital in the agriculture sector. Farmers are likely to make suchi investmentsmore rapidly in responseto improved incentives where there is a good rural infrastructure,secure property rights, strong public and private institutions, and well functionilng input, factors, and product markets and wdhere appropriatetechnologies are available.

SeeRepublic of iYemen: Ditnensions of Economic .4dijuxstmenciand .Structural Reform for a more detaileddiscussion on civil servicereform issues. - 28 -

3.25 Lessons from other low-income countries show that rural poverty can be significantly reduced through higher agriculture yields and development expenditures in the rural sector.3 Thus, in order to reduce poverty in Yemen, the Government will need to take initiatives which will raise the profitability of agriculture sector activities and thereby improve living standards of the majority of the population. Specific areas where the Government can intervene are:

* Eliminate the wheat and flour subsidy to remove the bias against the domestic production of grains and cereals. Since output responds strongly to price, grain production should increase. Also, because the growth in one crop usually takes resources away from others, food production may replace the production of qat.

* Since output is also affected by investments in roads, markets, and irrigation, infrastructure needs to be provided in the rural sector where it is efficient. This does not mean that there should be a rural bias, but rather that the Government should seek ways of encouraging investment in the rural areas where the returns are comparable to those achievable in other areas.

* Current rural practices such as depletion of water resources and deforestation call into question the sustainability of agriculture in Yemen. Therefore, policies and incentives need to be implemented which will reverse these processes. The World Bank is working with the Government to develop and implement a water strategy which should address the problems resulting from the unconstrained exploitation of nonrenewable water sources and growing monopolization in rural areas of water resources by better off farmers.

- Research and extension improve inputs into farm production such as fertilizer and seeds which in turn will increase the output of the agriculture sector.

See Ravallionand Datt. "Growvthand Poverty in Rural India".

See Republic of Yernen: Agriculture Sector Study - Strategy',Jbr Sustainable Agricultural Production, The World Bank. - 29 -

4. LONG-TERM IMPLICATIONS OF SOCIAL SECTOR POLICIES ON THE POOR

4.1 Long-term issues in economic development focus primarily on human resources. Since the objective of economic growth is social development, the sectors which play the most important role in improving living conditions are health and education.

4.2 The development of human resources in Yemen is limited. Mortality rates are high and life expectancy is low. Though recent investment in education has increased, literacy and enrollment rates continue to be low when compared to other low-income countries.

4.3 The state of human resources development in Yemen can be explained by several factors. First, since the population is dispersed (urbanization is about 20 percent), provision of services is difficult and expensive. Second, the Government is the major provider of social services, but does not have the financial, human capital, or institutional capacity to deliver a basic package. Third, consumer demand for health and education services is low for various reasons, including quality of services supplied, cost of services, individual preferences, education levels, and income levels.

4.4 Though there are many obstacles to wider coverage of the population, provision of services could be expanded under different incentives. This section proposes that the Govermment should ensure that the population (including the poor) has access to:

- A package of basic health services * Primary and secondary school education

Figure 4.1: Social Indicators

Under-5mortality

Infant mortality . l

Illiteracy j' YLIC n MENA Primary enrollment I

Life expectancyV l

0 20 40 60 80 100 120 140

Source: 11orld Development Report (1995). Note: (I) LIC is low-incomecountries. (2) MENA is Middle Eastand North Africa low andmiddle-income countries. (3) Under-5 mortality and infant mortality aredeaths per 1000children. -ife expectancyis in years. (4) Primary schoolenrollment and adult illiteracy arein percent. - 30 -

4.5 Though these goals appear straightforward, they imply new strategies in health and education and different financing and delivery mechanisms. One possible set of targets is presented below in Table 4.1. According to this vision for human resources development, enrollments and basic health coverage can be expanded by almost 50 percent and 125 percent respectively by the end of the decade such that school enrollments increase from 51 percent to 75 percent and basic health coverage grows from 20 percent to 45 percent. This expansion of services would require additional public funding equivalent to about $396 million or 6 percent of GDP. Since public expenditures are already too high, this expansion of social services can be funded by savings from the decrease in the petroleum and wheat subsidies.

4.6 The Government can use these as possible goals or outline its own based on more sector specific information. The following sections discuss ways to improve the health and education status of Yemen's population.

THE HEALTH SECTOR

4.7 Yemen's health status is characterized by high infant, child, and maternal mortality and low life expectancy. The leading causes of death in infants and children are diarrhea and acute respiratory infections (ARI). Additional significant health problems affecting children are protein- energy and deficiencies in iron, iodine, and vitamin D. Immunization coverage is low, less than 45 percent of 12 - 23 month old children. Life expectancy at 51 years is below average when compared to that of low income countries (62 years).

4.8 Yemen's total fertility rate of 7.5 births per woman is one of the highest in the world. Current child-spacing practices adversely affect the health of women and children. Poor women are more vulnerable than nonpoor women because they tend to have more children. The median birth interval is 27 months with particularly low intervals (20 months) for the high risk group of women under 20 years of age. This short birth-spacing is related to low contraception use, about 7 percent among women of childbearing age.

4.9 Since data on the health status of individuals by poverty status is unavailable, correlates of poverty (such as area, governorate, and literacy level of mother) can be used to obtain some information. Recent analysis on health conditions has shown that the infant mortality rate (considered a good indicator of a population's overall health status) is higher in northern governorates, the rural sector, and amongst infants born to mothers Nvhoare illiterate or have not received antenatal or delivery care during pregnancy.2 The two most vulnerable groups are children and women of child-bearing age. Children under the age of 5 years and women of child- bearing age constitute 16 percent and 17 percent of the poor respectively.

4.10 Most of the development of Yemen's public health system occurred after 1970. The system assumed several levels of services from health care units (949 in 1992) and centers (321 in 1992) at the primary level, to small and medium-size hospitals at the secondary level (87 hospitals), and a few large specialized (tertiary care) hospitals in the main cities.

Yemen: Demographic and Maternal and Child Health Survey 1991/92.

2 Yemen: Demographic and AtIaternaland Child Health Survey 1991/92. - 31 -

Taba 4.t1: Invesoment in Health and Education: Priori6es to Reduce Poverty

BASE PROJECTIONS Unit 1994 1996 1997 1998 1999 200 Ex-penditurerequirements GrandTotal nominal YR million 16,848 60,548 78,647 97,754 I118,158 143,739 Health3 nominal YR million 4,483 12,460 17,166 21,858 26,860 35,40 Education nominal YR million 12,365 48,088 61,481 75,896 91,297 108,330 rand Total nominal $ million 291 404 496 554 616 687 Health nominal $ million 78 96 III 129 147 181 Education nominal S million 214 370 399 446 501 554 As share of GDP percent 6.4 9.1 9.5 10.3 11 2 12.2 Health percent 1.7 1.9 2.1 2.3 2.5 3. Education percent 4.7 7.2 7.5 8.0 8.6 9.2 What it buys umber of people Covered by basic health package million 3.00 4.04 5.01 6.03 7.11 9.17 Of school age enrolled in school million 2.60 3.05 3.46 3.91 4.39 4.91 Share of population Covered by basic health package percent 20 25 30 35 40 50 6-18 years enrolled in school percent 51 55 60 65 70 75 Additional resources required nominal $ million 113 205 263 325 39

Where to get the moneyfrom |Total nominal S million 113 205 263 325 396 Wheat subsidy removal nominal $ million 63 130 163 200 246 Petroleum subsidy removal nominal $ million 50 75 100 125 150 Memorandum items: GDP at market prices nominal YR million 262,713 664,570 824,457 945,157 1,056,845 1,177,618 GDP at market prices nominal $ million 4,545 5,112 5,352 5,560 5,801 6,027 Population million 15.02 16.17 16.69 17.22 17.77 18.34 Population 6 - 18 years million 5.09 5.54 5.77 6.01 6.27 6.54 Avg. recurrent cost ofeduc. per student nominal YR million 4,723 11,251 13,281 14,597 15,649 16,639 Avg. capital cost for additional students4 nominal YR million 10,553 14,311 17,003 19,700 22,80 Avg. cost of basic health care package US$ per capita 12.0 12.7 13.1 13.5 13.9 14.3

Source: World Bank staff calculations based on information from Ministries of Finance, Education, and Health.

This scenario assumes that in 1994 57 percent of health care budget (S44 million) is not allocated to basic health services. For 1996 - 2000, this amount is assumed to grow at a rate of 3 percent per annum. Thus, in theory it is still possible to achieve efficiency gains, however given entrenched spending patterns this is unlikely to occur in the next few years. 4 The cost per classroom is assumed to be $8,000 in 1994 with a 3 percent increase every year. The student to classroom ratio is 35:1. If this increases, the cost of total capital investment will decrease accordingly. - 32 -

MAJORISSUES

4.11 Government expenditures on health are too low to provide the basic, minimum package of services. The World Development Report 1993 on health recommends a basic package of public health and clinical services with an estimated cost of $12 per person. This package is tailored to low-income countries such as Yemen. The level of financing suggested by the WDR indicates that the Government's expenditure on health ($5.6 per capita)' should be twice the current amount if all resources were used in the provision of basic health care. This would raise the share of the health budget allocation to I I percent of total expenditures and 3.7 percent of GDP.

4.12 The basic package of health services has two components. The public health component includes such items as EPI Plus,6 school health (i.e., deworming. micronutrient supplementation, and health education), information on health and nutrition, and vector control. The second component - an essential package of cliical services - includes treatment for , sick child management, prenatal and delivery care, and limited care for the treatment of common diseases (i.e., assessment, advice, alleviation J.fpair, treatment of infection, and minor trauma).

4.13 Reallocation of the health budget could lead to better services. Currently 66 percent of the Government's health budget is ailocated towards wages and salaries, however once the central ministry's budget is excluded, leaving only governorate expenditures, the share increases to 91 percent. The proportion of non-wage operating expenses is very low, and has fallen in the last four years. The situation is particularly serious in ten governorates where over 90 percent of the budget is allocated to wages leaving few resources for the purchase supplies such as drugs and x- ray films (which are supposed to be allocated centrally but are seldom delivered in sufficient quantities).

4.14 Poor governorates do not receive their share of funds. Governorates with a larger share of their population living in the urban sector receive more expenditures per capita than those with predominantly rural populations. This difference may be due to both the greater provision of services as well as the presence of more expensive structures (such as hospitals) in urban areas. Expenditures per capita vary greatly among governorates. For example, recurrent expenditures per capita in Aden are 50 times Saadah's.

4.15 The southern governorates in general have much higlherexpenditures which is a reflection of the emphasis placed on social development in the former PDRY. The health sector in the southern governorates is more developed and geared towards providing essential services to a large share of the population. Due to recent difficulties the health system has deteriorated, though the physical infrastructure remains for service delivery.

Calculatedusing the exchangerate for GDPconversion. At the parallel exchangerate expenditures measuredin US$ are lower. 6 ExpandedProgram for ImmunizationPlus which refersto a programfor increasingimmunization coverageof under-fivechildren as well as distributingadditional items such as vitaminA capsules. - 33 -

4.16 If out-of-pocket health-related expenditures increase, then the level of povertv would increase as well. The poverty line is calculated based on the assumption of a specific profile of seeking treatment. For example, the average number of diarrhea episodes suffered by children under-5 years is 7.3 episodes per year in the rural areas, 31 percent for which treatment is actually sought.7 If instead treatment was sought 80 percent of the time for the two major causes of death in young children (diarrhea and ARI), the level of poverty would increase from 19 percent to 45 percent due almost entirely to the increase of poverty in the rural sector from 19 percent to 50 percent. Thus, under the present conditions, better health will come at a price for the household. (See Annex V.A.)

Figure 4.2: Health Expenditures and Poverty by Governorate

35 1000 30 rV Healthexpenditures inAden 900 werew YR 995per cap ita 800 2 25 700 Ka v 20 600 P | 400 L---eat I o ~~~~~~~~~~~~~~~~~~~300 200 0C 5 1'00

C~ c . - < E Um

-~~~~~ ~~-

Source: World Bank staffcalculations based on HBS-92 and data from Ministry of Finance. Note: The governorates are ordered by southern and northem areas and then by incidence of poverty.

4.17 The cost of health care is particularly expensive to the rural population by a factor of about 3 compared to urban residents. The major difference in the cost per episode treated is the transportation and accommodation costs associated with seeking treatment in the vicinity or the nearest city. This is due to the still higher cost of providing health services in remote ruralareas. The decrease in the petroleum subsidy, though necessary for economic development, will adversely affect the cural population's (financial) access to health care.

4.18 The utilization of health services is low, particularly given the high rates of fertility and child morbidity. Utilization of health services can help to improve the health status of a population. In the case of Yemen where healti) conditions are inadequate, thq primary reasons for the low utilization of services are (i) limited availability of services, especially in the rural areas; (ii) limited access to available services, especially in teims of financial affordability; (iii) low quality of services; and (iv) limited availability of drugs.

Yemen: Demographic and Maternal and Child Health Survey 1991/92. - 34 -

4.19 Quality of health services in Yemen is low because of the limited clinical skills of medical staff and the lack of laboratory and x-ray facilities - tools useful in diagnosing ill health. Poor quality dissuades people from using services. However, when families do use health facilities, they frequently bypass local facilities to travel to a town where a greater range of services is available. The poor are less likely to be able to afford health services and are therefore more likely to experience higher mortality rates than the nonpoor.

4.20 Low availability of pharmaceuticals is a particularly important problem in public health services, as much of modem curative medicine depends on drugs. The management of medical supplies (i.e., procurement, storage, distribution, and utilization) is poor and most public facilities receive insufficient drugs8 . As a consequence, most drugs are smuggled into Yemen and are of variable - frequently low -quality.

4.21 Illiteracy among mothers is expensive in terms of child health, child survival, and family health costs. The consequences of a mother's illiteracy are borne by the children. Literate mothers have fewer children, who get sick less frequently, and they obtain proper treatment for their children more often than illiterate mothers. Illiteracy among mothers of young children is twice as prevalent among the poor than the nonpoor. 4.22 Health education helps parents take better care of their children, more economically. For example, a public health campaign advertised the use of oral rehydration solution as a method for treating diarrhea would increase the number of episodes treated, but at a fraction of the cost of a treatment in a health facility.

OPTIONS

4.23 The lack of services is greatest among persons living in the rural sector and urban slums, both areas with a high incidence of poverty. The Government can effectively help the poor by (i) improving access to and quality of public health services, (ii) improving the supply of drugs, (iii) rationalizing the role of the public and private sector in delivery of health services, and (iv) intersectoral initiatives to improve the general health environment. The implementation strategy should give particular attention to the role of the government in the health sector, the private sector, and how their activities can complement one another. This improvement can only be achieved by increasing public expenditure on health and reallocating resources from personnel to supplies and between different levels of health services.

4.24 The Government needs to formulate a health policy which specifically addresses how basic public health and clinical services of a minimum quality will be delivered to the rural sector and to low income groipa.

* Define the package of essential services to be delivered to the population. This should include basic public health services (such as immunization, health education), basic clinical services and referral facilities, standard treatment protocols with emphasis on treatment of the sick child and basic services related to reproductive health (e.g. antenatal, delivery and postnatal care);

8 In those facilitiesvisited during September1994 supplies had been adequatefor only 4 of the last 14 months. - 35 -

* Determine how this will be delivered to the rural population and the poor. This will require clarifying the roles of public and private providers and addsessing the issue of cost recovery; and

* Improve services and assure that this quality is maintained.

4.25 Public expenditures on health should be increased over the next five years to improve health care coverage. Overall public expenditures are inadequate for ensuring the supply of even basic public health and clinical services. The current system does not address the needs of the low income groups who, because quality of services is most inadequate in rural areas, bypass peripheral facilities at considerable expense. The priority for increased public expenditure on health services should go to:

* Improving provision of and accessibility to basic health services;

* Improving the quality of basic health services, especially primary health care and basic referral services. This will include improved pre-service and in-service training and supervision of primary health care workers; and

- Better preventive (particularly immunization and health education) services through training and provision of adequate resources such as vaccines.

4.26 The poor can be given basic health services by involving the private sector and nongovernmental organizations. The Government ias seeii its role as regulating, financing, and providing health care services rather than shar;ng the burden and responsibility with the private and non-profit sectors. Despite considerable investment in basic infrastructure, the current system does not deliver basic health services to a significant share of the population, especially to rural residents. Expanding the government health infrastructure in rural areas is not a solution since it would be too costly to provide a health facility for each of the 35,000 hamlets and villages, whose average population size is about 350 persons.

4.27 NGOs are experimenting with possible solutions, including cost recovery schemes which will promote the financial sustainability of the programs. An additional option, as yet unexplored, is to mobilize the private sector to deliver essential clinical services and selected public health interventions, particularly in rural areas. It is critical that these innovations are evaluated in economic as well as health terms in order to assess their relevance to efficient improvement of health services and the living standards of the poor.

4.28 The Government should formulate a new pharmaceuticals policy which removes the state monopoly on procurement. The existing policy which gives the Ministry of Health a monopoly on drug procurement leads to too few legal import of drugs. Why this occurs is unclear due to the lack of transparency of the procurement regime. Resources for pharmaceuticals appears to be diverted from the purchase of drugs for public facilities, thus leaving many public health clinics and hospitals without adequate supplies. As a result smuggled drugs of variable quality now constitute the majority of domestic supply. There is an urgent need to implement a policy which increases the supply of essential drugs to rural areas and the urban poor, possibly through liberalizing pharmaceutical procurement and increasing quality control of drugs imported. - 36 -

THE EDUCATIONSECTOR

4.29 The Government of Yemen has shown a strong commitment to education. Between 1972 and 1990, primary school enrollment increased from 30 percent to 76 percent respectively and secondary school enrollment from 5 percent to 3 1 percent. An indication of the Government's continued commitment to education is that it allocates significant public resources to this sector. In 1994 primary and secondary education expenditures were 15 percent of total public expenditures or 5 percent of GDP.

4.30 The majority of resources are allocated to basic education (grades I - 9). In 1993/94 of the 2.6 million general education students 93 percent were enrolled in basic education. This level utilizes about 90 percent of general education teachers, schools, and classrooms. The average number of students per class is 33 and 42 at the basic and secondary levels in 1994. The student to teacher ratios are 34:1 and 23:1 respectively.

4.31 Despite the significant share of public resources allocated to education, the quality of services is low. This is apparent from the high repetition rates in primary schools which are 35 percent and 20 percent for boys and girls respectively. The estimated length of time needed to complete primary education is 13 years, that is, double the prescribed six school years. Additional indicators of poor quality are the inadequacy of facilities. For example, 49 percent of basic education schools have no electricity, 47 percent no water, and 44 percent no toilets.'' The situation in secondary schools, though relatively better, is also difficult.

MEAJORISSIJES

4.32 Female enroilment rates are too low. In 1992 for every 100 boys enrolled in school, there were only 39 girls. Even though it is not uncommon in developing countries for fewer girls to go to school than boys, Yemen's gender gap is particularly large. For example, in developing countries on average for every 100 boys enrolled in primary education, there were 86 girls (in 1990). In Yemen, for every 100 boys in primary school, there were only 41 girls - less than half the international figure.

4.33 Disaggregated data by rural and urban areas provides a possible explanation for the wide gender difference in enrollments between Yemen. Though female enrollment is low relative to males, the difference is much greater in rural areas. In urban areas the female enrollment rate is 80 percent that of Males, whereas in rural areas it is 35 percent. (See Table 4.2.) Some governorates in particular have exceptionally low female enrollment rates in rural areas. In Shabwa, Hajjah, Dhamar, and Saadah, rural female enrollment rates are less than 10 percent.

9 Source: World 7'ahles1994, WorldBank. I0 RecentlyLhe Government of Yemen has changed from a primary/secondaryschool system to one with basic and secondarylevels. Basic school iFfrom grades I - 9 ane secondaryschool includesgrades 10 - 12. The Education Statistics 1990/1991, Ministryof Education. - 37 -

Table 4.2 Enrollment Rates by Area and Sex (in percent)

Male Female Total Rural 54 19 37 Urban 59 47 53 All Yemen 55 25 41 Source: World Bank Staff calculations based on HBS-92.

4.34 In Yemen, as in many countries, under-investment in girls' education is in part due to the inability of school systems to satisfy parental preferences. In Yemen 78 percent of parents in rural areas of northern governorates agreed that it was important to educate their daughters - nevertheless female enrollment is low. Parents who do not send their female children to school cite the following reasons: lack of separate and independent girls' schools,12 shortage of female teachers, unsuitability of school hours to the assigned time for housework, long distance from home (especially for girls over ten years old), and early marriage.

4.35 Secondary school enrollment rates are low for both boys and girls. In 1992 only 41 percent of all children between the ages of 6 and 18 were enrolled in school. This is due not only to low female enrollment, but also to low secondary school enrollments. For example, in 1994 though basic enrollment rates for boys were 103 percent, they were 32 percent for secondary school. Similarly for girls, the corresponding rates are 41 percent and 8 percent.'3

4.36 Though data are unavailable which would allow us to answer the question of whether poor students are less likely to enter secondary school than nonpoor students, governorate-level data provides some insight. Analysis shows that there is little correlation between the incidence of poverty in a governorate and either primary or secondary enrollment levels as seen in Figure 4.3. Thus, according to aggregate data, access to education does not appear to be related to the level of poverty in a governorate. (See Annex V.B for enrollment levels by governorate.)

4.37 Education expenditures have declined rapidly since unification. During 1990 - 1994 total expenditures decreased by 27 percent from YR 6,543 million (US$ 294 million) to YR 5,810 million (US$ 214 million).'4 Capital expenditures on education have decreased from 8 percent of total education expenditures to 2 percent in 1994 even though student enrollments have risen during this period by 28 percent. Thus, it appears that the Government has accommodated these additional students by increasing the number of students per classroom.

12 During 1992/1993about 77 percent of basic education schools were co-educational,21 percent were boys' schools,and 2 percentgirls, schools. In Aden, Lahij, and Al-Maharathere were no girls' schools. 3 Even though the basic enrollmentrate appears high for males, high repetitionrates imply that there are a large numberof enrolledstudents beyond the age group 6-14 years. The figurefor 1994 is in constant 1991 YR. In nominalprices the educationbudget was YR 12,365 million. US$ amountsare calculatedusing the GDP conversionexchange rate. - 38 -

Figure 4.3: Enrollment and Poverty by Governorate

18 r 16 31 14[

1 2

.~~~~~~~~. 10 ~~~~~~~~~~~~~~~20 8 15 a '~~~~~~~~~~~~~~~~~~~~~.~~~~Secondarv 8 15~~~~~~~~~~~~~~~~~~~~~ Povertv 60 O 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I0 4 5. 2 V 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~0

mC > -< '0 S '0 -< uG E '

Source: World Bank Staffcalculationsbased on IBS-92 and datafrom Ministry of Education..

4.38 Total expenditureper studenthas fallen by 43 percentto $84 per student in 1994.15 A more appropriatemeasurement for how much the Government is spendingon educatinga child is the recurrentexpenditure per studentwhich decreasedto YR 4,723 (or US$ 82) in 1994. Though thereare no quantitative indicatorsavailable which would allow measuringquality changesduring this time period, it is probable that this significant decline in financial resourceshas had a large impact on the quality of public educationservices.

4.39 By 1994the total educationbudget was the recurrentbudget with 82 percentof it allocated to wagesand salariesand 13 percentto goods and services(chapter 2 of the budget). However, even though capital expenditureswere almost entirely curtailed, this did not serve to protect real expenditurelevels in the recurrentbudget.

4.40 Educationexpenditures per studentvary greatlyby governorateand are not correlatedwith the incidenceof poverty. During 1990- 1994,expenditures in absoluteterms have beenfallen by 27 percent. In 1994expenditure per studentwas highestin Shabwaat YR 14,502per studentand lowest in Sana'a City at YR 2,697 -a multiple of five times. Both of these governorateshave less than average poverty. The allocation of the education budget is correlated with the governorate'spopulation (rather than the student population) and with regional variables(i.e., north/south)as seenin Figure4.4.

This wascalculated at the exchange rate for GDPconversion. If theparallel market exchange rate were used,expenditure per studentwould have decreased during 1991 - 1994by 35 percent. - 39 -

Figure 4.4: Distribution of Population and Education Expenditures byvGovernorate

6 14 Hadramout lias 5%ofYemen's population

10 andrecieves 10%of the educationbudgt

Population

~6 i; - s F Z Exp:Education 4

'.< E

Source: World Bank staff calculationsbased on IIBS-92 anddata from Ministr of Finance.

4.41 Though their appears to be a degree of equity in allocation of budgetary resources to each governorate, the variation in expenditure per student may result in differences in the quality of education services provided unless it reflects the variation in the cost per student (due perhaps to economies of scale, wage supplements to teachers in rural areas, etc.). However, it is more likely that the expenditure per student reflects the difference in inputs per student and indeed in the quality of education services.

4.42 Inputs into education appear low in some governorates. especially those with high 16 enrollments. The average number of students per teacher decreased from 36 to 27 during 1990 - 1994. The number of students per classroom was about 34 in 1994. Though these ratios appear to indicate that inputs per student are reasonable, there are significant variations among governorates. For example, in Sana'a City (where a relatively large share of the school-age population is enrolled) the number of students per classroom in basic education is 73 compared to 14 students in Marib. A similarly large gap exists for the student to teacher ratio. Saadah has the highest student to teacher ratio with 91 studenits per teacher at the basic level compared to 17 in Abyan. For secondary school, these ratios also vary widely among governorates.

4.43 The student to input ratios are not correlated with the incidence of poverty in the governorate. Resources, such as teachers and classrooms, do not appear to be allocated on the basis of enrollments. Areas with a high demand for schooling appear to be penalized because on a per student basis they receive fewer inputs.

16 Data on education inputs are limited, therefore the discussion focussesprimarily on classrooms and teachers. - 40 -

Table 4.3 Cross-country Comparison of Enrollment and Public Expenditures on Education

Year GDP/capita Education Exp. Education Exp. Education Exp. Enrollment Rates in GDP in Total Exp. per student Primary Secondary (US$) (in percent) (in percent) (US$) (in percent) (in percent)

Yemen 1992 340 6.0 15 80 76 37 Low income countries 1992 .. . 14 .. 101 42

Malawi 1988 178 1.7 2 26 66 4 Madagascar 1991 233 1.8 17 22 92 18 Kenya 1992 315 5.9 22 88 95 29 Gambia 1990 338 1.8 11 41 69 18 Ghana 1992 431 6.0 24 135 74 38 Egypt 1992 762 2.2 6 55 101 80 Source: GovernmentFinance Statistics (IMF) and W1orldDevelopment Report 1995. Note: (1) Education expenditures are for preprimary, primary, and secondary levels. Data for Ghana, Kenya, and Malawi education expenditures include university expenditures as well. (2) Education expenditure per student are estimated based on the assumption that 6 - 18 year old children compose 38 percent of the population.

4.44 Yemen's education system is inefficient relative to other comparator countries. In 1992 the Government spent 6 percent of GDP on education, yet enrollments are relatively low when compared to other low-income countries. This seems to indicate that perhaps the allocation of resources within the sector could be improved. There is too little information to allow understanding the source of the inefficiencies, however, as a first step the allocation between teachers and administrators and other non-teaching personnel should be reviewed.

4.45 Out-of-pocket expenditures on education can be prohibitive for the poor.17 One estimate of the expenditure per student per year is YR 520 for poor families, YR 803 for middle class families, and YR 1,521 for affluent families. Given the change in prices and the exchange rate, it is estimated that the 1995 cost ranges between YR 2,500 to YR 8,300 per student per year. Since families on average have 3 school-age children, the cost becomes restrictive, particularly for the poor who have 23 percent more children. Given such costs, the problem of improving the educational profile of the poor needs to take into account the financial capabilities of the poor families to take advantage of the government willingness to expand such facilities.

7 The mere cost of transportationto a school could be prohibitiveto a poor family. On visiting a vocationaltraining center in the outskirtsof Sana'a, it was found that commutingto and from the center could cost aroundYR 50 daily (or YR 1300per month per student). la Hodeidah Standards of Living Survey (1994). - 41 -

OPTIONS

4.46 The Government should focus on increasing female enrollments. The Government may want to work with communities to develop interventions which would increase the proportion of girls that enroll in school. Possible interventions may be (i) construct culturally appropriate facilities with latrines for girls and possibly with walls around girls' classrooms; (ii) increase the number of female schools or have an additional shift only for girls; (iii) launch information campaigns that engage community, religious, and civic leaders; (iv) provide (financial) incentives for female teachers; (v) adjust the school calendar and schedule to accommodate household child labor requirements: and (vi) provide families with incentives for sending their girls to school (such as paying for female students' school materials and uniforms).

4.47 The Goverinment should establish basic school enrollment targets for the next five years. Current policy as well as expenditures in the education sector do not appear to be geared towards systematically expanding education opportunities for the population, whether poor or nonpoor. Without setting goals and knowing the financial cost of achieving them, it is difficult to ensure the implementation of any strategy. The Government needs to determine the present enrollment rates and decide what are reasonable goals for the next five years perhaps drawing upon the experience of other countries.

4.48 The quality of education can be improved through more classrooms and teachers, and better teaching. The lack of education inputs has many negative effects which include (i) poor quality of human capital produced, (ii) high dropout rates, (iii) high repetition rates which increases the burden on existing resources, and (iv) the high cost of expenditure per student. In order to achieve expansion in enrollments, this lack of resources for expanding coverage of the school-age population needs to be addressed. Student-teacher and student-classroom ratios in urban areas need to be brought down to acceptable levels no higher than 45 students per teacher and per classroom.

4.49 Yemen has a shortage of teachers who are adequately qualified, willing to teach in the rural sector, and are female. One possible solution to this supply-side constraint is for the Government to establish a program which would attract foreign male and female teachers to work in the under-served rural and urban areas. Since the living conditions will probably be less than they desire, a wage supplement will have to be paid to attract qualified personnel. A second possible solution to increase the number of teachers is expansion of enrollment in colleges of education, recruitment of university graduates fro,n non-education specializations, and enhancing teacher training capacities. Third, the Governrn-.et miy wvantto -onsider working with the private education sector such that private schools would deliver services to the under-served areas and the Government would finance their operating expenditures.

4.50 Community involvement should be increased, especially in school construction. co- financing of teachers, and quality control of teachers. The cost of classroom construction can slow rapid expansion of enrollment. One estimate for classroom cost (building, furniture, and equipment) is about $8,000 (YR 0.9 million) or $121 million per year to increase student enrollment by 500,000 students at the current student-classroom ratios. If the community donates the classrooms, this will save time and money. Greater involvement by the community in hiring and financing teachers may result in more quality control as parents' own resources are being used - 42 - to obtain a service. Even poor communities can contribute by providing labor if the government provides materials.

4.51 The Government can increase enrollment rates of poor children by finding ways to decrease the private cost of education. One of the major expenses for the poor are expenditures on school uniforms. During the period where the Government is making a concerted effort to increase enrollments, it may want to relax the school uniform requirement or subsidize uniforms in poorer regions. Other costs such as textbooks and stationery are also burdensome. One Yemeni NGO has taken the initiative of providing school supplies to poor children. This system could be expanded to cover more schools by encouraging NGOs to link themselves to specific schools in poor areas. They could also provide school supplies or provide scholarships to poor students.

4.52 The Government should establish a data and information system. The education sector has weak data and information, making it difficult to manage and evaluate. For example, even basic data on enrollment rates are unavailable. In this area the sector requires support on two levels, the first is to enhance the capacities of the Planning and Finance division at the Ministry of Education. The second is to provide support for Educational Research and Development Council within the context of a reorganization of its purpose and mandate. The establishment of an information and data support system (including school mapping) is vital for sector development and for policy implementation leading to an improvement of the educational situation. - 43 -

5. CONCLUSION

5.1 The strategy for poverty reduction in Yemen has three major components: (i) economic growth, especially in the labor-using tradeable goods sector, (ii) improved delivery and expansion of effective health and education services to the population, particularly in rural areas, and (iii) development of the nongovernmental sector. The strategy does not include the development of a publicly administered or financed social safety net.

5.2 Three characteristics shape the strategy for poverty reduction in Yemen: (i) the high incidence of poverty in the rural sector, (ii) the multi-dimensional nature of poverty which encompasses both low expenditures and lack of access to basic services, and (iii) the low capacity of the public sector to deliver assistance to the poor.

5.3 In order to reduce poverty the Government must be committed to revisiting the distribution of public expenditures, its social sector strategies, and its role vis a vis the private and nonprofit sectors. These steps are important for helping the existing poor and for stemming the growth in poverty. The actions proposed in this report not only support broad-based economic growth since it is a prerequisite for poverty reduction in Yemen, but also go beyond it to support rapid social development.

5.4 The analysis of the Government budget highlights two significant points. First, expenditures on health and education are not related to the level of poverty in a governorate - thus, need is not the criteria for distribution of public resources in the social sectors. Second, the large inequality between expenditures per capita by governorates indicate that there exist significant differences in the absorptive capacities of governorates, the efficiency with which they use resources, or in their living conditions.

5.5 The low social indicators in Yemen signify that the population as a whole suffers from lack of access to essential health and education services. Though data are not available by income group, all evidence suggests higher incidence of ill health and mortality among the poor than the nonpoor. In recent years the gap has narrowed between the educational attainment of the poor and nonpoor but remains large between males and females. For poverty reduction to occur, the divergence between male and female literacy rates needs to be eliminated, if only, because of the impact a mother's education has on family health and child school enrollment.

5.6 Though the problems are clear cut, the solutions require a comprehensive rethinking of the way the Government delivers social assistance to the population. It is clear that the Government should not finance universal subsidies which benefit the nonpoor more than the poor. It is also apparent that the Government is unlikely to possess the financial or administrative capacity to target cash or in kind transfers to the majority of the poor. The alternative is for the Government to ensure increased provision of those services which improve the poor's living conditions and encourage the development of the nongovernmental sector in providing assistance to the poor. - 44 -

5.7 Yemen faces many problems which the Government cannot solve alone. Given the extent of absolute poverty, the Government needs to reevaluate its role in the three step process of service provisioni: regulation, finance, and delivery. For Yemen to close the widening gap between itself and other Middle East countries, the Government has to find private sector counterpartsin the delivery of social services. Though the private and nongovernmental sectors are not well-developed, they can act as complements to the public sector giving the Government opportunities to pilot various solutions.

5.8 In order to ensure that the poor do not get bypassed in efforts to improve the delivery of assistance to the population, the Government must ensure that the services provided are the type which the poor need, can afford to purchase, and can access (geographical proximity).

5.9 Though poverty in Yemen is widespread, the Government can help to decrease it. Over the next several years oil revenues will continue to be significant and can be used by the Government to invest in the development of social services and physical infrastructure. This complemented by policies which encourage private investment and, therefore, job creation, should result in raising the quality of life and decreasing the number of poor. - 45 -

ANNExI.A

STATISTICAL BASIS FOR POVERTY MEASUREMENT

DATA SOURCES

1.1 The Household Budget Survey (HBS), on which the analysis in this report is based, contains information on 60.550 individuals and 8,405 households for 1992. The number of poor in the sample are 11,058. Data was collected on personal (such as age, sex, and marital status), household, and labor force characteristics as well as on household expenditures and individual income.

1.2 The survey is not self-representative but over-samples urban households and particular governorates. For example. urban observations comprised 72 percent of the sample though the share of the urban population is estimated at only 20 percent. The resulting biases were corrected using weights derived from the Statistical Yearbook 1993 and preliminary census data for 1994. The weights were derived using two sets of assumptions. First, the urban-rural ratio for each governorate was based on the census data collected in 1986 and 1988 for northern and southern governorates respectively. Second, governorate weights were calculated using preliminary results of th-ie Deceinber 1994 census provided by the Central Statistical Organization.

1.3 In order for the survey to be nationally representative, a minimum number of observations are required per governorate and in each urban and rural area. In certain categories the number of observations appear low may which may call into question the reliability of the information from that particular area.

1.4 The poverty profile excluded three households (or 31 observations) whose expenditures per month per capita were in excess of YR 50,000. These households appeared to be outliers since the households with total expenditures closest to them were about YR 36,000 per month per capita.

1.5 Throughout this report an individual's or household's living standard is measured using expenditures rather than income. This is the common approach taken in poverty analysis for two reasons. First, reported income is unreliable due to the tendency of people to understate their earnings. Second, consumption is a better indicator of welfare because households can protect their living standards from income fluctuations to some extent. Since expenditures are reported at the household level only, total expenditures are divided by the number of household members to obtain per capita figures. Household expenditures include the imputed value of consumption in kind, such as from own-consumption.

THE POVERTY LINE

1.6 Four poverty liies were calculated for each governorate: a lower and upper poverty line for the urban area and another set for the rural area. The lower (food) poverty line is equivalent to the amount of money needed to purchase 2200 calories; the composition of the food bundle and the retail prices used to calculate the cost of the bundle are outlined in Table I.B. Certain modifications were made as can be seen by comparing the FAO food category to the actual commodity prices. For example, red meats were substituted with bread beans and maize and millet with imported white wheat. Though substitutions were made, the caloric intake of the new item was the same as that of the original food item. - 46 -

1.7 The upper poverty line is the lower poverty line plus a nonfood component. The equation used for calculating the higher poverty line is:

Upper Poverty Line = Lower Poverty Line * (2 - x) where x is the share of food in total expenditiu-es c' individua,ls with total mean expenditures equal to the lower poverty line. This mean s the average of the food t3 tota; expenditure ratios of individuals with expenditures in increments of YR 5 atound the inean.

1.8 Since the Department of Statistics collects retail prices for only twelve of the eighteen governorates, poverty line estimates for rural areas and the remaining governorates were based upon the following assumptions. First, governorates' poverty lines were calculated using average regional prices. For example, since retail prices for Abyan were lacking, the average of Aden's and Al-Beida's prices were used. Second, since no rural prices were collected, they were assumed to be 15 percent below urban prices. This estimate of the rural/urban differential is based upon data from other developing countries. (See Table I .C.)

Governorateweights used in the povertyprofile calculations

Governorate Weights Number of Observations Urban Rural Urban Rural Total

lbb 3.74 46.97 2,670 2,371 5,041 Abyan 5.50 44.27 696 500 1,196 Sana'aCity 4.96 0.00 13,171 0 13,171 Al-Beida 6.06 60.62 622 464 1,086 Taiz 3.71 44.49 5,137 2,711 7.848 Al-Jawf 0.62 52.31 547 199 746 Hajjah 3.91 57.66 853 1.413 2,266 Al-Hodeidah 4.26 47.48 6,967 1,624 8,591 Hadramout 3.78 41.79 4,206 797 5,003 Dhamar 6.11 43.04 825 1,444 2,269 Shabwa 7.20 68.31 429 313 742 Sadaah 10.49 52.96 258 569 827 Sana'a 6.28 59.14 575 2,092 2,667 Aden 5.27 4.43 5,159 677 5,836 Lahij 4.99 55.14 538 687 1,225 Marib 2.31 35.38 282 304 586 Al-Mahwit 4.17 41.04 272 595 867 Al-Mahara 6.29 7.59 247 305 552

All Yemen 43,454 17,065 60,519

Source: World Bank staff calculations based on HBS-92 and Statistical Yearbook. Data for the Calculationof the Poverty Linefor 1992

FAO food Consumption Actual Average Retail Prices for 1992 (in fils) category Calories kg/year commodity Ibb Hodeidah Taiz Aden Sana'a Al-Beida Hajjah Dhamar Al-Jawf Marib Al-Mahwit 817 Wheat 782 97.7 impornedwhite wheat 725 618 700 358 800 833 870 817 967 833 Rice 72 10.7 cheapestrice 1375 1510 1233 1075 1375 1391 1510 1400 1629 1510 1422 Barley 54 5.9 barley 1592 1542 1292 1504 1525 1627 1218 1608 1945 1773 1630 Maize 57 6.4 imported white wheat 725 618 700 358 800 833 870 817 .Y67 833 817 Millet 25 2.9 imported white wheat 725 618 700 358 800 833 870 817 9S7 833 817 Sorghum 437 51.7 red 1517 1229 1290 1452 1525 1550 1527 1525 1563 1600 1583 Potatoes 56 27.3 1642 1364 1567 1253 1642 1583 1318 1583 2000 1767 1792 Sugar 261 26.5 1517 1427 1656 1350 1517 1567 1518 1567 1759 1658 1567 Pulses 55 5 9 chick peas 4292 4333 4750 4337 4842 5458 5500 5883 6000 5950 4900 tomatoes 7 15 5 1625 1392 1292 1304 1609 1725 1745 1683 2000 1767 1792 onions 3 39 white onions 1750 1518 1667 1497 1917 1833 1591 1875 2208 1833 1727 veg, other 8 8.1 cabbage 1292 1571 1542 1296 1208 1475 1875 1200 1875 1639 1600 oranges 1 0.9 3100 3500 3300 3250 3000 3300 3450 3000 3000 3000 3167 lemons, limes 0.5 lemons 5542 '1C50 5938 3738 5227 5750 6600 6000 6125 5625 6250 bananas 6 3.5 1458 1630 1608 1438 1458 1500 1773 1375 1732 1593 1542 .&, apples 0 0.2 red apples 6500 8000 5583 6281 7500 6375 5000 6500 C625 5500 5333 pineapples 1 06 bananas 1458 !63, 1608 1438 1458 1500 1773 1375 1792 1593 1542 ' dates 8 2.0 freshdates 5250 3681 4136 4615 5500 5167 6750 5292 5682 5333 5417 grapes 22 12.2 black grapes 2250 2250 2850 2422 2100 2167 1375 2417 2400 2100 2250 fruit,others 20 34.8 watermelon 1142 1313 1818 747 1108 1008 1517 1150 650 592 1133 coffee 1 0.5 yemeni seeds 19111 20333 15727 18279 20200 20000 17500 19111 24000 21333 18667 tea 1 07 lipton 4750 6111 4667 5850 4833 4750 4955 4917 5500 5182 4875 spices 0.2 salt 417 660 375 243 410 492 473 403 500 492 480 poultry 35 10.1 perchicken 9293 6364 10167 8242 9042 9208 9318 8583 10833 10500 9208 meatothers 1 0.3 beef 10833 117.-7 11500 13756 10750 11000 13636 10833 11292 11833 11333 milk 77 115.4 liquid milk 1/2 It 1064 lOIS 1050 910 1073 1140 982 1060 1500 1200 1010 eggs 9 1 6 dozen 3300 3009 3100 3120 3250 3350 3518 3250 3750 3350 3300 fish 7 3.6 9125 SC45 11000 6330 8842 9292 7133 9083 88X2 8842 9625 cooking oil 119 1 4 cookingoil chief 9292 8836 9125 8824 9083 9333 8636 9458 10375 9458 9417 butter/ghee 24 1.2 ghee 3517 2945 3125 2990 3492 3542 3400 3517 3800 3558 3525 fats, other 10 0.4 ghee 3517 2945 3125 2990 3492 3542 3400 3517 3800 3558 3525 red meats 58 5.9 bread beans 3833 3250 4583 4012 3417 3542 3370 3625 4389 4333 3773

Source: FAO Food Balance Sheets for Yemen Arab Republic (1984 - 1986) and Statistical Yearbook 1993. Note: (1) Entries in italics signify that prices were missing and therefore regional averages were used instead. (2) Red meats were subsituted for equal caloric value with beans since the latter is a less expensive nutriitonal substitute. - 48 -

ANNEX I.C

Upperand Lower PovertyLines for 1992by Governorate (in YR per capita per year)

URBAN RURAL Govemorate Lower Upper Lower Upper lbb 12,974 16,528 11,028 14,048 Abyan 13,294 18,951 11,299 16,109 Sana'a City 12,763 19,759 -N/A- -N/A- Al-Beida 13,951 25,200 11,858 21,420 Taiz 12,787 18,174 10,869 15,448 Al-Jawf 15,791 21,433 13,422 18,218 Hajjah 13,775 22,437 11,709 19,071 Hodeidah 12,254 17,331 10,416 14,732 Hadramout 13,127 17,642 11,158 14,996 Dhamar 13,625 18,580 11,581 15,793 Shabwa 14,645 20,119 12,448 17,101 Sa'ada 13,548 19,721 11,516 16,763 Sana'a Govemorate 12,763 17,523 10,849 14,894 Aden 12,636 17,577 10,741 14,940 Lahij 13,087 21,741 11,124 18,480 Marib 15,338 25,153 13,037 21,380 I-Mahwit 14,035 18,864 11,930 16,035 Al-Mahrah 13,127 26,255 11,158 22,317

Weighted National Lower Poverty Line 11,496 Weighted National Upper Poverty Line 16,488

Source: World Bankstaff calculationsbased on HBS-92and StatisticalYearbook 1993. - 49 -

ANNEX IlA

Summaryof Poverty Indices

Govemorate Population Distribution Average Expenditure Headcount Index Poverty Gap Squared Poverty Distribution of the Poor of Poor Index Gap Index (in percent) (in percent) (YR/Year/Person) (in percent) (in percent) (in percent)

Ibb 12.1 12.2 5,969 19.1 5.0 2.0 Abyan 2.6 2.0 6,718 14.7 3.5 1.4 Sana'aCity 6.5 3.7 8,745 10.9 2.6 1.1 Al-Beida 3.2 5.6 9,169 33.2 7.5 2.6 Taiz 14.0 13.5 6,608 18.4 5.7 2.5 Al-Jawf 1.1 <.1 9,439 0.3 0.1 0.0 Hajah 8.5 4.4 8,125 10.0 2.2 0.7 Hodeidah 10.7 10.1 6,319 18.0 4.4 1.7 Hadramout 4.9 5.7 5,582 21.9 8.1 4.5 Dhamar 6.8 11.9 5,350 33.6 13.0 7.0 Shabwa 2.4 0.9 7,560 7.1 1.3 0.3 Saadah 3.3 0.6 7,726 3.8 0.4 0.1 Sana'a Gov 12.7 17.5 5,784 26.2 8.7 3.9 Aden 3.0 2.8 5,628 17.8 6.0 2.9 Lahij 4.1 3.7 7,179 17.2 5.1 2.3 Marib 1.1 0.7 7,832 12.5 3.6 1.2 Ai-Mahwit 2.6 4.3 5,607 32.0 11.3 5.8 Al-Mahara 0.4 0.4 11,508 17.9 3.4 0.9

All Yemen 100.0 100.0 6,435 19.1 5.7 2.6 Source:World Bankstaff calculations based on HBS-92. - 50-

ANEX ILB

Level of Educationby PovertyStatus and Sex (18years and above)

EducationalLevel PM Nonpoo All Yemen Women Men Women Men Men Women

Illiterate 85 44 83 34 35 83 Literate 7 29 8 32 31 8 Primary 2 7 3 8 8 3 Preparatory 2 10 3 10 10 3 Secondary 1 2 1 2 2 1 Communitycollege 2 6 3 11 10 2 Post-secondaryinst. <1 I

Total 100 100 100 100 100 lOO

Source:World Bankstaff calculationsbased on HBS-92. - 51 -

ANNEX II.C

Enrollment Rates(%) y total cohotp 50 - 60 / F- 60 - 70 //\

UADAH~~~~~~ ALMAHARA \

i / ~~~~~~~~~HADRAMO T

AL PAe e'@ HBt

HODEI D H

Source: World Bank Staff Calculations based on HBS-92.

The map shows enrollment rates of children between 6-18 years by governorate. The denominator is the governorate's total cohort population. - 52 -

ANNEX II.D

Employment Sector and Poverty

Unpaid 40% of the poor are self-emp loy ed. Employer

Self-employed !Nopoor

Private Sector

PublicSector

0 5 10 15 20 25 30 35 40 (in percent)

Source: World Bank Staff Calculations based on HBS-92. - 53 -

ANNEX 11E

Determinants of Expenditures

Parameter Probability Weighted Standard Variable Estimate T-Statistic T = 0 Mean Deviation

Intercept 9.52 75.78 0.00 Householdsize -0.02 -3.59 0.00 7.18 14.74 Sexof headof household 0.03 0.60 0.55 0.97 0.74 Age of headof household 0.00 0.64 0 52 42.92 54.65 Ageofheadofhousehold squared 0.00 -0.56 0.57 2024.26 5138.71 Literatespouse 0.11 4.16 0.00 0.10 1.21 Headof householdwith primaryschool 0.13 3.29 0.00 0.04 0.83 Headof householdwiith mtermediateschool 0 11 2.64 0.01 0.04 0.81 Headof householdwith secondaryschool 0.10 1.17 0.24 0.01 0.36 Headof houseoholdwith communitycollege 0.20 5.21 0.00 0.05 0.B9 Headof householdwith postsecondary certificate 0.09 1.10 0.27 0.01 0.38 Headof householdwri- universitydegree 0.33 5.96 0.00 0.02 0.60 Headof householdwit:. postgraduate degree 0.36 1.88 0.06 0.00 0.18 Headof householdin public sectoremployee 0.26 4.20 0.00 0.20 1.62 Headof householdin privatesector employment 0 13 2.19 0.03 0.17 1.52 Headof householdis employer 0 43 5.43 0.00 0 02 0.54 Headofhouseholdworkswithoutp.y 0.12 1.33 0.18 0.01 0.41 Headof householdis self-employed 0.24 4.00 0.00 0.52 2.02 Headof householdis retired 0.13 1.78 0.07 0.02 0.49 Headof householdreceives remittances 0.38 5.52 0.00 0.06 0.99 Headof householdis married 0.07 1.83 0.07 0.90 1.21 Headof householdis single 0.01 0.10 0.92 0.05 0.89 Headof householdlives in urbanarea 0.00 0.12 0.90 1.80 1.62 Numberof otherpublic sectorworkers in household 0.04 2.41 0.02 0.17 2.19 Numberof otherprivate sector workers in household 0.00 0.49 0.62 0.51 4.02 Numberof childrenunder six yearsof age -0.07 -7.64 0.00 1.38 5.71 Numberof childrenbetween seven and twelve years -0.05 -6 12 0.00 1.49 5.67 Numberof childrenbetween thirteen and eighteen years -0.05 -5.64 0.00 1 25 5 46 Abyan 0.34 6 31 0.00 0.02 0.62 Sana'aCity 0.54 12.83 0.00 0.07 1.00 Al-Beida 0.40 8 02 0.00 0.03 0.65 Taiz 0.35 12.31 0.00 0.14 1.41 Al-Jawf 0.90 12.45 0.00 0.01 0.42 Hajjah 0.52 15 79 0.00 0.08 1.12 Hodeidah 0.15 5.17 0 00 0.13 1.37 Hadramout 0.06 1 44 0.15 004 0.77 Dhamar -0.07 -2.07 0.04 0.07 1.01 Shabwa 0.45 8.34 0.00 0 02 0.59 Saadah 072 15.17 0.00 0.03 0.70 Sana'aGov -0.03 -0.96 0.31 0.11 1.28 Aden 0.04 0.87 0.39 0.03 0.71 Lahij 0.33 7.31 000 0.03 0.74 Marib 0.44 6.08 0.00 0.01 0.42 Al-Mahwit -0.21 -4.25 0.00 0.03 0.63 Al-Mahara 0.70 5.85 0.00 0.00 0.25

DependentVariable: Log of annualper capitahousehold expenditures Meanof dependentvariable 9.79

AdjustedR square 0.23 Source:World Bank staff calculationsbased on HBS-92. - 54 -

ANNEX 1II

Distribution of Expenditures in Yemen

Expenditure per capita Share of Population Cumtilative Share of Population (YR per month) (in percent) (in percent) <100 0.2 0.2 100-200 0.7 (.X 200-300 1.3 2.1 300-400 1.9 4.0 400-500 3.8 7.8 500-600 4.0 11.8 600-700 4.5 16.3 700-400 5.8 22.1 800-90( 5.0 27.1 900-10(0 6.3 33.4 1100-1200 5.4 38.8 1200-1300 4.7 43.5 1300-1400 5.3 48.7 1400-1500 4.7 53.4 1500-1600 5.0 58.4 1600-1700 3.5 61.9 1799-1800 3.1 65.0 18(0-1900 3.1 68.1 1900-2000 2.8 70.9 2000-2100 3.1 74.1 2100-2200 1.9 76.0 2200-2300 1.9 77.9 2300-2400 2.0 79.8 2400-2500 1.8 81.6 2500-2600 1.5 83.2 2600-2700 1.1 84.3 2700-2800 1.4 85.7 2800-2900 1.4 87.0 2900-3000 1.0 88.0 3000-3100 0.8 X8.X 3100-3200 0.7 89.5 3200-3300 0.8 90.4 3300-3400 0.5 90.9 3400-3500 0.6 91.4 3500-3600 0.5 91.9 3600-3700 0.4 92.4 3700-3800 0.4 92.8 3800-3900 0.4 93.2 3900-4000 0).6 93.8 4000-4100 0.5 94.3 4100-4200 0.1 94.5 4200-4300 0.2 94.7 4300+ 5.3 100.0

Source: World Bank stafTcalculations based on HBS-92. - 55 -

ANNEX IV.A

RegionL Defeh.se Expettdil;ircs

D)efenseE pend. DefenseExpenditures DefenseExpenditures GNP Country Year in total c\penditures per capita as shareof GNP per capita (in percent) (in tUS$) (in percent) (in lIS$)

Y'emen 1994 34 351 112 3272 IJAF 1993 3X 928 4 21,430 Onmani 1993 35 993 20 4,850 Jordan 1992 2'2 82 7 1,120 Isracl 1993 20 1.178 8 13,920 Ku%%ait 1993 210 2.063 11 19,360 [3ahrain 1993 i h 465 Fgypt 1992 X 25 4 660 Iran 1993 , 18 1 2,200

Source World Llankstaff calculationshased on GovernmentFinance Statistics Yearbook. World Tables 1994, and UWorltL)eve!opment Report. Nole: ( I ) Calculatedat the exchangerate for (il)P conversionwhich is YR 57.8/USS. (2) As shareol GDP sincereliahle (iNP data is unavailable. - 56 -

ANNEXIV.B

PublicSector Employment Change 1991 1994 (in percent) Public sector employment (in '000) 389.7 596.7 53 o/w civilian 193.9 291.0 50 o/w military 144.7 218.7 51 o/w public enterprises 51.1 87.0 70

Population (in '000) 13400 14300 Labor force/population (%) 23.8 23.8 Labor force size (in '000) 3,189 3,403

Govt employees/labor force (%) 12.2 17.5 Govt employees/population (%) 2.9 4.2

Wage Bill (YR ml nominal) 25.7 49.8 o/w civilian 14.7 28.4 o/w mIito'ry 11.0 21.4

Wage Bill /Shareof GDP) 23.0 19.0 -17 o/w civilian 13.1 10.8 -18 o/w military 9.8 8.1 -17

Ave.age Civilian Wages at current prices (YR '000) 75.9 97.7 29 at current prices (US$) 2,988 1,216 -59

|Memorandumitems:l GDP mp (YR ml nominal) 111,965 262,713 Parallel Exchange Rate (YR/USS) 25.41 103.00

Source: Republic of Yemen: Dimensions of Economic Adjustment and Structural Reform and Ministry of Finance. - 57 -

ANNEx V.A

Estimated Annual PrivateCosts of Treatmentfor Diarrheaand ARI (per child under 5 years)

Share of episodes Episodes Episodes for which for which Cost per Treatment per child care sought care sought episode costs Area Illness Prevalence per year treated per year (in percent) (number) (in percent) (number) (YR) (YR) lJrban Diarrhea 26.3 4.2 58.9 2.5 940 2325 ARI 17.7 2.8 71.0 2.0 940 1869

Rural Diarrhea 36.0 7.3 30.9 2.3 2625 5921 ARI 26.1 4.1 31.8 1.3 2625 3422

Source: World Bank staff calculations based on Yemen: Demographic and Maternal and Child Health Survey 1991/92. - 58-

ANNEXV.B

EducationIndicators by Governorate

Governorate Adult Illiteracy Gross Enrollment Male Female Total Male Female Total Ibb 30 91 60 65 31 50 Abyan 22 52 3 7 55 44 50 Sana'a City 14 64 38 61 49 56 Al-Beida 26 84 55 59 20 42 Taiz 30 86 58 53 29 42 Al-Jawf 33 90 64 42 34 39 Hajjah 64 95 79 38 7 25 Hodeidah 54 88 70 51 24 39 Fladramout 22 57 40 58 30 45 Dhamar 47 94 70 53 8 32 Shabwa 31 91 60 49 8 30 Saadah 47 92 69 52 11 34 Sana'a Gov 37 91 64 58 16 37 Aden 13 43 28 54 52 53 Lahij 16 73 45 60 21 39 Marib 39 84 62 37 18 28 Al-Mahwit 36 98 66 56 13 38 Al-Mahara 23 60 41 54 39 45

All Yemen 35 83 59 55 25 41

Source: World Bank staff calculations based on HBS-92. - 59-

REFERENCES

As-Suswa, Amat Al-Alim. 1994. Democratic Development in Yemen. USA.

Binswanger, Hans. 1990. "The Policy Response of Agriculture." In Proceedings of the World BankAnnual Conference on Development Economics 1989. USA. pp. 231-257.

Faini, Riccardo and Jaime de Melo. 1993. Fiscal Issues in Adjustment in Developing Countries. USA.

Food and Agriculture Organization. 1991. Food Balance Sheets: 1984-86Average. Italy.

Government of Yemen. 1991 - 1995. Statistical Yearbook. Yemen.

Government of Yemen. 1993. The Education Statistics Yearbook 1990/91. Yemen.

Government of Yemen, Pan Arab Project for Child Development, and Macro International Inc. 1994. Yemen: Demographic and Maternal and Child Health Survey 1991/92. Yemen.

Grosh, Margaret E. 1994. Administering Targeted Social Programs in Latin America: From Platitudes to Practice. USA.

International Monetary Fund. 1994. Government Finance Statistics Yearbook. USA.

Jongstra, E. 1994. Hodeidah Standards ofLiving Survey. Yemen.

Ravallion, Martin. 1994. Poverty Comparisons. USA.

Ravallion, Martin and Benu Bidani. 1994. "How Robust is a Poverty Profile?" In The World Bank Economic Review vol. 8(1):75 - 102. USA.

Ravallion, Martin and Gaurav Datt. 1995. "Growth and Poverty in Rural India". Policy Research Woilking Paper, no. 1405, World 3ank. USA.

Sud, Inder. 1995. "Infrastructure Investment in the Middle East." World Bank. USA.

UNDP. 1995. Human Development Report. USA.

UNICEF and the Government of Yemen. 1993. The Situation of Children and Women in the Republic of Yemen: 1992. Yemen.

World Bank. 1994. Republic of Yemen: Agriculture Sector Study: Strategy for Sustainable Agricultural Production. USA.

World Bank. 1995. Republic of Yemen. Dimensions of Economic Adjustment and Structural Reform. USA.

World Bank. 1993 - 1995. Wor:dDevelopmentReport. USA.

World Bank. 1994. World Tables 1994. USA.

MAP SECTION

REPUBLIC OF YEMEN POVERTY ASSESSMENT PROJECT

- GOVERNORATE BOUNDARIES 00 b° - 1

INTERNATIONAL BOUNDARIES

* NATIONAL CAPITAL ,

| A)AAA / A MH4 \

HA5ORAMOL'r

''P\~~~~~~~~iI IA WJ'P' I / 1/. f.l 'IA .,JAH ) |Aj /

SA NA 'A , /

*AL MAHWT vSa a City HARWA

.~~~~~~~~~~~~LHOD)EIPWLiH BDA

\ _ ~~~~~~~~~~~~4 BYAN\ N~~~~~~~ w LAHIJ / Socotra (Yemen)

'Abd cl Kuri - ADEN (Yemen)

- Sormhah Dorsch

(The Brothers) (Yemen)

m~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-kl- -. -y*<

ffi t_K m - , _ -f e tho

- ° ~~~~~~~~~~~~~~~~~~1,c) E inc~~~~~~~~~b2 0 0 4-1d.*. 0 i

REPUBLIC OF YENIEN

PRIORITYOP()VER I NDI(CA1ORS

Unit 1960 1970 1980 1990 1992

Social Intlicators Infant Mortality Rate per 1000live births 210 192 143 124 118 Female-to-MaleLife Expectancy years 36.35 41/41 46/44 49/48 51/50 Total Fertility Rate childreni 7.6 7.6 7.5 7.5 Maternal Mortality per 1(00,000live birihs .. .. 330

Gross Primary Enrollienit percenit 1 22 51 .. 76 Adult Illiteracy percent ...... 62

1990 1991 1992 1993 1994

Ek-onioiyii Intlicators Governmentsocial expenditures o/w educationexpenditures perceit of GDP 5.1 5.8 6.0 6.1 4.7 o/v healthiexpenditures percentof GDP 1.4 1.8 1.7 1.8 1.7

GDP percapita lJSSper vear 578 476 337 297 307

GDP nomilinalLUS$ ml 6.868 5,038 4.674 4.254 4,545 GDP nominal YR ml 96-367 111,965 134,708 169.793 262,713

Averageofficial exchangerate Y'R/US$ 11.07 12.01 12.01 12.01 12.01 Parallelexchange rate YR/US$ 14.59 25.41 33.38 48.73 80.34

Population millions 11.9 13.4 13.9 14.3 14.8

Source: World Developnment Repiort(various years), Ministriesof Finance,Education. and Health. iMAG INCA

Report No: 15158 YEM Type: SR